Embed Size (px)
DESCRIPTION
Alice tier 1 testbed in KISTI-NSDC. 2010. 7. 8 Christophe BONNAUD and Beob Kyun KIM, NSDC / KISTI. 1. Contents. Introduction to KISTI Introduction to NSDC Project Activities in 2009 System architecture Management plan for Alice tier-1 testbed Expected technical issues. - PowerPoint PPT Presentation
Citation preview
Alice tier 1 testbedin KISTI-NSDC
1
2010 7 8Christophe BONNAUD and Beob Kyun KIM
NSDC KISTI
1WLCWLCG workshopG workshop
Contents
Introduction to KISTI
Introduction to NSDC Project
Activities in 2009
System architecture
Management plan for Alice tier-1 testbed
Expected technical issues
2WLCG workshop
Introduction to KISTI
WLCG workshop
CDF Offline Meeting FNAL
SEOUL
Daejeon
160km
KISTI
bull 6 Universitiesbull 20 government re-
search institutesbull 10 government-invested institutesbull 33 private RampD labs bull 824 high-tech compa-
nies
Located in the heart of Science Valley ldquoDaedeokrdquo In-nopolis
The DAEDEOK INNOPOLIS complex consists of a cluster of firms that represents a cross-section of Koreas cutting-edge industries including information technology biotechnology and nanotechnology
a Government supported Research Instituteto serve Korea Scientists and Engineers with ScientificIndustrial
Data Bases Research Networks amp Supercomputing Infrastructure
WLCG workshop
We are the pioneer of building Cyber-Infrastructure of Korea
SampT InformationResource amp Service
Technological amp IndustrialInformation Analysis Supercomputing service
High PerformanceRampD Network service National
Cyber RampD Infrastructure
NationalRampD Promotion
Annual revenue about US-D100M mostly funded by the government
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Contents
Introduction to KISTI
Introduction to NSDC Project
Activities in 2009
System architecture
Management plan for Alice tier-1 testbed
Expected technical issues
2WLCG workshop
Introduction to KISTI
WLCG workshop
CDF Offline Meeting FNAL
SEOUL
Daejeon
160km
KISTI
bull 6 Universitiesbull 20 government re-
search institutesbull 10 government-invested institutesbull 33 private RampD labs bull 824 high-tech compa-
nies
Located in the heart of Science Valley ldquoDaedeokrdquo In-nopolis
The DAEDEOK INNOPOLIS complex consists of a cluster of firms that represents a cross-section of Koreas cutting-edge industries including information technology biotechnology and nanotechnology
a Government supported Research Instituteto serve Korea Scientists and Engineers with ScientificIndustrial
Data Bases Research Networks amp Supercomputing Infrastructure
WLCG workshop
We are the pioneer of building Cyber-Infrastructure of Korea
SampT InformationResource amp Service
Technological amp IndustrialInformation Analysis Supercomputing service
High PerformanceRampD Network service National
Cyber RampD Infrastructure
NationalRampD Promotion
Annual revenue about US-D100M mostly funded by the government
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Introduction to KISTI
WLCG workshop
CDF Offline Meeting FNAL
SEOUL
Daejeon
160km
KISTI
bull 6 Universitiesbull 20 government re-
search institutesbull 10 government-invested institutesbull 33 private RampD labs bull 824 high-tech compa-
nies
Located in the heart of Science Valley ldquoDaedeokrdquo In-nopolis
The DAEDEOK INNOPOLIS complex consists of a cluster of firms that represents a cross-section of Koreas cutting-edge industries including information technology biotechnology and nanotechnology
a Government supported Research Instituteto serve Korea Scientists and Engineers with ScientificIndustrial
Data Bases Research Networks amp Supercomputing Infrastructure
WLCG workshop
We are the pioneer of building Cyber-Infrastructure of Korea
SampT InformationResource amp Service
Technological amp IndustrialInformation Analysis Supercomputing service
High PerformanceRampD Network service National
Cyber RampD Infrastructure
NationalRampD Promotion
Annual revenue about US-D100M mostly funded by the government
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
CDF Offline Meeting FNAL
SEOUL
Daejeon
160km
KISTI
bull 6 Universitiesbull 20 government re-
search institutesbull 10 government-invested institutesbull 33 private RampD labs bull 824 high-tech compa-
nies
Located in the heart of Science Valley ldquoDaedeokrdquo In-nopolis
The DAEDEOK INNOPOLIS complex consists of a cluster of firms that represents a cross-section of Koreas cutting-edge industries including information technology biotechnology and nanotechnology
a Government supported Research Instituteto serve Korea Scientists and Engineers with ScientificIndustrial
Data Bases Research Networks amp Supercomputing Infrastructure
WLCG workshop
We are the pioneer of building Cyber-Infrastructure of Korea
SampT InformationResource amp Service
Technological amp IndustrialInformation Analysis Supercomputing service
High PerformanceRampD Network service National
Cyber RampD Infrastructure
NationalRampD Promotion
Annual revenue about US-D100M mostly funded by the government
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
a Government supported Research Instituteto serve Korea Scientists and Engineers with ScientificIndustrial
Data Bases Research Networks amp Supercomputing Infrastructure
WLCG workshop
We are the pioneer of building Cyber-Infrastructure of Korea
SampT InformationResource amp Service
Technological amp IndustrialInformation Analysis Supercomputing service
High PerformanceRampD Network service National
Cyber RampD Infrastructure
NationalRampD Promotion
Annual revenue about US-D100M mostly funded by the government
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
History of KISTI Supercomputers
1988 1993 20001997 2001
Cray-2S1st SC System
in Koreauntil 1993
Cray C902nd SC System
in Koreauntil 2001
Cray T3Euntil 2002
HP GS320 HPC160320
NEC SX-5
IBM p690
CAVE ampSGI Onyx3400
2002
PC Cluster128node
SC 45
2003
IBM p690+37TF
NEC SX-6 TeraCluster
2 GFlops 115 GFlops 240 GFlops 285 TFlops
436 TFlops435 GFlops111 GFlops16 GFlops
2GF 16GF 242GF131GF 786GF 52 TF 86TF2007 200835TF 280TF
CDF Offline Meeting FNAL (090903)
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
WLCG workshop
World Class Facilities at KISTI
300
2560034
Ranked 15th in TOP500 2010
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
KREONET is the national sci-ence amp research network of Ko-rea funded by MOST since 198820Gbps backbone 1 ~ 20Gbps access networks
National Research Network
GLORIAD (GLObal RIng Network for Advanced Appli-cations Development) with 1040Gbps Optical lambda networkingndash Global Ring topology for advanced science applicationsndash GLORIAD Consortia Korea USA China Russia Canada
the Netherlands and 5 Nordic Countries (11 nations) Essential to support advanced application develop-
ments HEP ITER Astronomy Earth System Bio-Medical HDTV
etc National GLORIAD project (funded by MOST of KOREA)
CDF Offline Meeting FNAL (090903)
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Introduction to NSDC project
WLCG workshop 9
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
10
Introduction
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
WLCG workshop 11
Introduction
Data Intensive Researches
Bio-informaticsAstrophysics
Earth ScienceHigh Energy Physics
CDF Belle ALICE
2009 ~
Next ~
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
12
History
bull Planned by Governmentndash A project to build a strategic plan and a model of NSDC
bull MEST (Ministry of Education Science and Technology)bull 2008 02 ~ 2008 05bull ldquo 사이버 RampD 인프라 구축 활용방안 기획연구rdquo
bull Most important things for this project isndash The contribution to Korean researchersrsquo study
WLCG workshop
This project is a part of GOVERNMENTrsquoS MASTER PLAN
FOR SCIENCE AND TECHNOLOGY DEVELOPMENT ( 과학기술기본계획 2009~)
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
13
Activities in 2009
ALICE Tier2 CenterBelleCDF
DMRC amp Neuroimaging
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
KISTI ALICE Tier2
Resources
ndash 10 services for ALICE(WMS+LB PX RGMA BDII VOMS LFC VOBOX CREAMCE UI)
ndash 120 core 21888 kSI2kbull HP Blade (25GHz Quad 2) x 15 set
WLCG workshop 14
2008 2009 Provided Comments
CPU (kSI2K) 100 150 219 CREAM-CE
Storage (TB) 30 TB 50 TB 30 56TB Physical
Network (Gbps) 10 10 10
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
KISTI ALICE Tier2
WLCG workshop 15
Site Availability to EGEE983 (Feb2009~)
CREAM-CE support- First 3 sites in the world
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
NSDC system architecture
HardwareGrid Middleware Deployment
WLCG workshop 16
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
17
ResourcesCluster
Storage
Service nodes for Grid amp Application ~20 nodes (including VM)
WLCG workshop
Cluster Spec node nodes cores kSI2kce-al-ice
Dell Intel Xeon E5405 x2 (20GHz Quad) 16GB
6 48 48
ce01 HP Intel Xeon E5420 x2 (25Ghz Quad) 16GB
15 128 219
ce02 IBM Intel Xeon E5450 x2 (30GHz Quad) 16GB
38 304 650
Intel Xeon X5650 x2 (266GHz Hex) 24GB 36 432 --Total 114 976 --Model Capacity Physi-
callyCapacity Usable
NetApp FAS2050 (SAN only RAID6 HA) 48TB 30TBNetApp FAS6080 (SAN amp NAS RAID6 HA) 334TB 200TBHitachi (SAN only) + IBRIX 950TB 600TBTotal 13PB 830TB
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
VOBOXs
VOBOXs
Storage Pool
rsquo100702
ALICE
NSD
C
Computing Resourcesbull 60node (480core)
6 node 48 core- Xeon E5405 (20GHz) 2GB RAMcore- 1075 si2kcore AVG
DELL 1U (pe1950)WN
16 node 128 core- Xeon E5420 (25GHz) 2GB RAMcore- 1710 si2kcore AVG
HP BladeWN
10G 38 node 304 core- Xeon E5450 (30GHz) 2GB RAMcore- 2177 si2kcore AVG
IBM 1U (x3550)WN
px
CREAM-CE
lcg-CE
UIs
WMS+LB
RGMA
BDII-top
BDII-site
MyProxy
VOMS
VOBOXs
GUMS-OSG
CE-OSG
LFC
dgassdfarmkrNagios
repositorysdfarmkrRepository
SMURF
NetApp FAS6080Hitachi USP-V
IBRIX
Belle
LDG Server LIGO
SAM
36 node 432 core- Xeon X5650 (266GHz) 2GB RAMcore
IBM BladeWN
NetApp FAS2050
Belle db
bull 1349TB (830 TB usable)
Xrd head
Xrd head
SEDPM
Xrootd (t1)
Xrootd (t2)
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Resource Plan in 2010
WLCG workshop 19
600TB usable space is being configured (Total 830TB Usable) ALICE Belle LIGO STAR GBRAIN hellip
Storage
36 blade servers HS22 (Xeon X5650 6 core x2 RAM 24GB HD 146GB)
Computing
Between storage and computingbull Basically all computing resource have 1G dedicated con-
nectionbull To enhance access speed Parallel File System is adaptedbull Special nodes like storage will have 10G connection to
storage
Internal Networking
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Resource Plan in 2010
WLCG workshop 20U600TB
NAS NetAppFAS6080 A-1
NAS NetAppFAS2050 NAS NetApp
FAS6080 A-2
IBM server ldquox3550rdquo x 14
Dell 1U Server ldquoPE1950rdquo x 16
HP Blade ldquoBL460crdquo x 16
Y 2008 Install
IBM 1U server ldquox3550rdquo x 38
Y 2009 Install
IBM Blade Server ldquox5650rdquo x 36
Y 2010 Install
KREONET To Backbone Network
Public Network
Data Network
1Gb10Gb8Gb FC SAN
Intensive IO Servers
SMC8848M
Juniper EX2500
Cisco 3560E
HP IBRIX9300 (x5)
10 links4 links
HP StorageworksX9300(SAN Switch)
2 links
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Management plan for Alice tier-1 testbed
HardwareGrid Middleware Deployment
WLCG workshop 21
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Resources
WLCG workshop 22
Disk 100TB Managed by Xrootd
One new redirector Tape 100TB (Tape emulation on disk)
Storage
Shared resourcesComputing
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Human resources
WLCG workshop 23
14 Members but only 9 for IT 4 members affected to Tier-1 testbed
Network power management and aircooling managed by other teams
Actual status
1 more person with strong experience in Tier-1 management will be recruited from any other country
We will invite experts for Tier-1 management as many as pos-sible
Future
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Expected technical issues
HardwareGrid Middleware Deployment
WLCG workshop 24
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Resources
WLCG workshop 25
tape storage needed to become a real Tier-1 Which hardware Which software (price seems critical) Manpowerexpertise
Storage
How to deal with a single farm shared by many projects 2 VOBOXs with same VO (Alice Tier-1 and Tier-2) issues
queue name Env Variables (software storage ele-ment hellip)
How to control resources usages
Farm
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
Monitoring
WLCG workshop 26
What other sites use What is available
Existing Software
Nagios for real time SMURF for historic No production monitoring
Current status
Critical for timemanpower consumption Bugs hunt project on Mobile Phone development
Home made software
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
WLCG workshop 2727
General Contact nsdcnsdckr Speaker cbonnaudkistirekr
Project Leader Haengjin JANG hjjangkistirekr
httpnsdckr will be available in soon
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
KISTI Supercomputing CenterExtending the horizon of science and technology
ResourceCenter
TestbedCenter
Keep securingprovid-ing world-class super-
computing systemsResourceCenter
Help Korea research commu-nities to be equipped with proper knowledge of super-computing
TestbedCenter
Validate newly emerg-ing concepts ideas tools and systems
Make the best use ofwhat the center has to create the new values
TechnologyCenter
Value-addingCenter
National headquarters of Supercomputing re-sources e-Science Grid and high performance research networks
WLCG workshop
![KISTI 해수 담수화 기술.ppt [호환 모드]](https://img.pdfslide.tips/doc/110x75/61c6d0f9525ae2513553d505/kisti-ppt.jpg)
