View
704
Download
1
Category
Tags:
Preview:
DESCRIPTION
With the rapid growth of the production and storage of large scale data sets it is important to investigate methods to drive the cost of storage systems down. Manyenergy conservation techniques have been proposed to achieve high energy efficiencyin disk systems. Unfortunately, growing evidence shows that energy-saving schemes in disk drives usually have negative impacts on storage systems. Existing reliability models are inadequate to estimate reliability of parallel disk systems equipped with energy conservation techniques. To solve this problem, we firstly propose a mathematical model - called MINT - to evaluate the reliability of a parallel disk system where energy-saving mechanisms are implemented. In this dissertation, MINT is focused on modeling the reliability impacts of two well-known energy-saving techniques - the Popular Disk Concentration technique (PDC) and the Massive Array of Idle Disks (MAID). Different from MAID and PDC which store a complete file on the same disk, the Redundancy Array of Inexpensive Disks (RAID) stripes file into several parts and stores them on different disks to ensure higher parallelism, hence higher I/O performance. However, RAID faces more challenges on energy efficiencyand reliability issues. In order to evaluate the reliability of power-aware RAID, wethen develop a Weibull-based model–MREED. In this dissertation, we use MREED to model the reliability impacts of a famous energy efficiency storage mechanism– the Power-Aware RAID (PARAID). Thirdly, we focus on validation of two models–MINT and MREED. It is challenging to validate the accuracy of reliability models, since we are unable to watch certain energy-efficiency systems for a couple of decades due to its time consuming and experimental costs. We introduce validated storage systemsimulator–DiskSim–to determine if our model and DiskSim agree with one another. In our validation process, we compare a file access trace in a real-world file system. Last part of of this dissertation focuses on improvement of energy-efficient parallel storage systems. We propose a strategy–Disk Swapping–to improve disk reliability by alternating disks storing data that is frequently accessed with disks holding less accessed data. In this part, we focus on studying reliability improvement of PDC and MAID. At last, we further improve disk reliability by introducing multiple diskswapping strategy.
Citation preview
Reliability Modeling and Analysis of Energy-Efficient Storage Systems
Shu Yin
Advisor: Dr. Xiao QinCommittee Members: Dr. Sanjeev Baskiyar
Dr. Alvin LimUniversity Reader: Dr. Shiwen Mao
2
Presentation Outline
MotivationMINT ModelMREED ModelModels ValidationReliability ImprovementConclusion and Future Work
2
3
Motivation
Data Intensive Applications
Stream Multimedia Bioinformatic
3D Graphic
BioinformaticBioinformatic
Weather Forecast
Bioinformatic
4
Data Intensive Computing Application
Cluster System
5
Problem: Energy Dissipation
EPA Report to Congress on Server and Data Center Energy Efficiency, 2007
6
Problem:Energy Dissipation(cont.)
Using 2010 Historical Trends Scenario
Data Centers consume 110 Billion kWh per Year;
Assume Average Commercial End User Is Charged ¢9.46 per kWh
Disk System Can Account for 27% of the Computing Energy Cost of Data Centers.
Disk Syste
m27%
Other73%
Disk System May Have An Electrical Cost of
2.8 Billion Dollars!
7
Existing Energy Conservation Techniques
Software-Directed Power ManagementDynamic Power ManagementRedundancy TechniqueMulti- speed Setting
7
How Reliable Are They?
8
Contradictory of Energy Efficiency and Reliability
Example: Disk Spin Up and Down
Energy Efficiency
Reliability
9
Presentation Outline
Motivation
MINT ModelMREED ModelModels ValidationReliability ImprovementConclusion and Future Work
9
10
MINT(MATHEMATICAL RELIABILITY MODELS FOR ENERGY-EFFICIENT PARALLEL DISK SYSTEMS)
Energy Conservation Techniques
Single Disk Reliability Model
System-Level Reliability Model
11
Frequency Utilization
Disk Age Temperature
Reliability of Single Disk
Single Disk Reliability Model
MINT(Single Disk)
12
MINT(Single Disk)
R=α*BaseValue[1]*TemperatureFactor+β*FrequencyAdder[2]
α and β are two coefficients to R
Assumption: α = β = 1 in our research
[1] E. Pinheiro, W.-D. Weber, and L.A. Barroso. Failure trends in a large disk drive population. Proc. USENIX Conf. File and Storage Tech., February2007.
[2] IDEMA Standards. Specification of hard disk drive reliability.
13
MINT(Single Disk)
R=α*BaseValue*TemperatureFactor+β*FrequencyAdder
Utilization Impact on AFR
Temperature Impact on Temperature Factor
Transition Frequency Impact on Frequency Adder
14
MINT(Single Disk)
R=α*BaseValue*TemperatureFactor+β*FrequencyAdder
Single Disk Reliability
Frequency=250/Month, T=40°C
Frequency=350/Month, T=35°C
Frequency=250/Month, T=35°C
Base Value from Google Report[3]
[3] E. Pinheiro, W.-D. Weber, and L.A. Barroso. Failure trends in a large disk drive population. Proc. USENIX Conf. File and Storage Tech., February 2007.
Frequency=350/Month, T=40°C
15
MINT(Energy Conservation Techniques- PDC)
- hot data
- cold dataPopular Date Concentration (PDC)[3]
System Structure
[3] E. Pinheiro and R. Bianchini. Energy conservation techniques for disk array-based servers. Int’l Conf. on Supercomputing, pages 68–78, June 2004.
16
MINT(Energy Conservation Techniques- PDC)
More Popular Disk Less Popular Disk
Access Rate<MIN(Access Rate)
Access Rate<MIN(Access Rate)
Access Rate>MAX(Access Rate)
Access Rate>MAX(Access Rate)
- hot data
- cold data
17
MINT(Energy Conservation Techniques- PDC)
- hot data
- cold data
(Optimal Result for Certain Time Phases)
Popular Date Concentration (PDC)[3]
System Structure
18
MINT(Energy Conservation Techniques- MAID)
- hot data
- cold dataMassive Array of Idle Disks (MAID)[4]
System Structure
[4] Dennis Colarelli and Dirk Grunwald. Massive arrays of idle disks for storage archives. Supercomputing ’02: Proceedings of the 2002 ACM/IEEE conference on Supercomputing, pages 1–11, Los Alamitos, CA, USA, 2002. IEEE Computer Society Press.
19
- hot data
- cold dataMassive Array of Idle Disks (MAID)[4]
System Structure
[4] Dennis Colarelli and Dirk Grunwald. Massive arrays of idle disks for storage archives. Supercomputing ’02: Proceedings of the 2002 ACM/IEEE conference on Supercomputing, pages 1–11, Los Alamitos, CA, USA, 2002. IEEE Computer Society Press.
Access Rate>MAX(Access Rate)
Cache Disk Data Disk
MINT(Energy Conservation Techniques- MAID)
20
MINT(System-Level)
Energy Conservation Techniques
Single Disk Reliability Model
System-Level Reliability Model
Reliability of Disk 1
Reliability of Disk n
Frequency Utilization
TemperatureAccess Pattern
Frequency Utilization
Disk Age
Reliability of A Parallel Disk System
21
Preliminary Results(experimental setting)
Energy-efficiency Scheme
Number of DisksFile Access Rate(No. per month)
File Size(KB)
PDC20 data
(20 in total)0~106 300
MAID-115 data + 5 cache
(20 in total) 0~106 300
MAID-220 data + 5 cache
(25 in total) 0~106 300
Read-only Disks
22
Preliminary ResultComparison Between PDC and MAID
AFR Comparison of PDC and MAIDAccess Rate(*104) Impacts on AFR (T=35°C)
23
Preliminary ResultComparison Between PDC and MAID
AFR Comparison of PDC and MAIDAccess Rate(*104) Impacts on AFR (T=35°C)
- MAID- PDC
24
MAID under High Access Rate
MAID-1
MAID-2
AFR Comparison of PDC and MAIDAccess Rate(*104) Impacts on AFR (T=35°C)
25
MAID under High Access Rate
AFR Comparison of PDC and MAIDAccess Rate(*104) Impacts on AFR (T=35°C)
MAID-1
MAID-2
MAID-1
MAID-2
MAID-1
MAID-2
26
MINT(conclusion)
Mathematical Model for Disk Systems MINT Study on PDC and MAIDBut ...
What about RAID?Data Stripping Mechanism
Energy Consumption IssuesReliability Issues
Complexity
27
Presentation Outline
MotivationMINT Model
MREED ModelModels ValidationReliability ImprovementConclusion and Future Work
27
28
MREED Model(MATHEMATICAL RELIABILITY MODELS FOR ENERGY-EFFICIENT RAID SYSTEMS)
Access Pattern Temperature
Energy Conservation Techniques
Frequency
Utilization
Annual Failure Rate
Weibull Analysis
29
Weibull Analysis
A Leading Method for Fitting Life Date Advantages:
AccurateSmall SamplesWidely Used
29
30
MREED Model(Energy Conservation Techniques- PARAID)
SoftState
RAID
Gears
321
Power-Aware RAID (PA-RAID)[5]
System Structure
[5] Charles Weddle, Mathew Oldhan, Jin Qian, An-I Andy Wang.PARAID: A Gear-Shifting Power-Aware RAID. USENIX FAST 2007.
31
Reliability Evaluation(Experiment Setup)
Disk Type Seagate ST3146855FC
Capacity 146 GB
Cache Size Sata 16MB
Buffer to Host Transfer Rate 4Gb/s (Max)
Total Number of Disks 5
File Size 100 MB
Number of Files 1000
Synthetic Trace Poisson Distribution
Time Period 24 Hours
Interval Time (Time Phase) 1 Hour
Power on Hour Per Year 8760 Hours
32
Reliability Evaluation(Disk Utilization Comparison)
Disk Utilization Comparison Between PARAID-0 and RAID-0 at A Low Access Rate (20/hr)
33
Reliability Evaluation(Disk Utilization Comparison)
Disk Utilization Comparison Between PARAID-0 and RAID-0 at A High Access Rate (80/hr)
34
Reliability Evaluation(AFR Comparison)
AFR Comparison Between PARAID-0 and RAID-0 at A Low Access Rate (20/hr)
35
Reliability Evaluation(AFR Comparison)
AF
R
AFR Comparison Between PARAID-0 and RAID-0 at A High Access Rate (80/hr)
36
Presentation Outline
MotivationMINT ModelMREED Model
Models ValidationReliability ImprovementConclusion and Future Work
36
37
Model Validation
TechniquesRun the Systems for A Couple of Decades
The Event Validity Validation Techniques[6]
[6] R.G. Sargent, “Verification and Validation of Simulation Models”, in Proceedings of the 37 th conference on Winter Simulation, ser. WSC’05 Winter Simulation Conference, 2005.
38
Model Validation
ChallengesUnable to Monitor PARAID Running for Years
Sample Size is Small from A Validation Perspective (e.g. 100 Disks for Five Years)
39
Model Validation(DiskSim[7] Simulation)
[7] S.W.S John, S. Bucy, Jiri Schindler and G.R. Ganger, “The DiskSim Simulation Environment Version 4.0 Reference Manual”, 2008
File To Block Level Converter
40
Model Validation(DiskSim Simulation)
Diagram of the Storage System Corresponding to the DiskSim RAID-0
41
Model Validation(Result)
Utilization Comparison Between MREED and DiskSim Simulator
42
Model Validation(Result)
Gear Shifting Comparison Between MREED and DiskSim Simulator
43
Presentation Outline
MotivationMINT ModelMREED ModelModels Validation
Reliability ImprovementConclusion and Future Work
43
44
Recall PDC
- hot data
- cold data
(Optimal Result for Certain Time Phases)
Popular Date Concentration (PDC)System Structure
45
Problem of PDC
The Most Popular Disk:High AFRNo Replica
46
Reliability Improvement of PDC
Method of Improving ReliabilityMirroring
Extra Disks for Replication -> More Energy Consumption
Disk SwappingSwap Existing Disks
47
Disk Swapping SchemePDC
Swap the Most Popular Disk with the Least Popular Disk
48
Swap the Highest AFR Disk with the Lowest AFR Disk
Disk Swapping SchemePDC
49
Swap the Cache Disks with the Data Disks
Disk Swapping SchemeMAID
50
Preliminary Results(experimental setting)
Energy-efficiency Scheme
Number of DisksFile Access Rate(No. per month)
File Size(KB)
PDC20 data
(20 in total)0~106 300
MAID-115 data + 5 cache
(20 in total) 0~106 300
MAID-220 data + 5 cache
(25 in total) 0~106 300
Read-only Disks
Mean Time to Data Lose (MTTDL)
Swapping Thresholds (2*105, 5*105, 8*105 No./Month)
Single Swapping
51
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR
(T=35°C)Threshold = 2*105 No./Month
Comparison of Disk SwapPDC
52
Comparison of Disk SwapPDC
AFR:Swap2 < Swap1 < No Swap
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR
(T=35°C)Threshold = 2*105 No./Month
53
Comparison Between Different Threshold
PDC
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR
(T=35°C)Threshold = 2*105 No./Month
54
Comparison Between Different Threshold
PDC
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR
(T=35°C)Threshold = 5*105 No./Month
55
Comparison Between Different Threshold
PDC
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR
(T=35°C)Threshold = 8*105 No./Month
56
AFR Comparison of PDCAccess Rate(*104) Impacts on AFR (T=35°C)
Threshold = 2*105 No./Month, 5*105 No./Month, 8*105 No./Month
Comparison Between Different Threshold
PDC
AFRHigher Threshold -> Lower AFR
57
Limitations
Read Only Disk Scenario
Data Migration within Certain Time Phases
Simple File Access Patterns
58
Future Work
Extend the Models to investigate mixed read/write workloads;
Research the trade-offs between reliability and energy- efficiency;
Extend schemes to a real-world based environment;
Develop a multi-swapping mechanism
balancing the utilization & lowering the failure rate;
Evaluate more control groups.
59
Conclusion
Generic Models coupled with power management optimization policies;
Two reliability models for the three well-known energy-saving schemes -- PDC, MAID and PARAID;
Disk swapping strategies to improve disk reliability for PDC.
Thanks
Questions?
Recommended