Upload
mongodb
View
235
Download
1
Embed Size (px)
Citation preview
1
NetApp + MongoDB: Maximize Your Investment
MongoDB Europe 2016
Paul Mu Technical Director, NetApp
Mohinder Toor Business Development Executive, NetApp
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Agenda1. NetApp Introduction
2. MongoDB Internal Storage Pain Points
3. Why NetApp
4. MongoDB Use Cases: NetApp Optimizes your MongoDB solution
High Performance (AFF)
Cost-Effective Performance (EF)
Enterprise Ready Infrastructure (AFF)
Eliminate Data Sprawl and Reduce Data Copies (AFF)
Best Breed Data Life Cycle Management (AFF, SolidFire)
5. Resources
6. Q&A
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only2
3
1. From FORTUNE Magazine, June 15, 2016 © 2016 Time Inc. FORTUNE and FORTUNE 500 are registered trademarks of Time Inc. and are used under license. FORTUNE and Time Inc. are not aff iliated w ith, and do not
endorse products or services of, NetApp.IDC Worldw ide Quarterly Enterprise Storage Systems Tracker 2016 Q1, June 2016 (Open Netw orked Enterprise Storage Systems revenue)
in Converged
Infrastructure
capacity shipped
FY’16 revenue:
$5.55
billion
Recognized as a
FORTUNE 500® Company
20161
Creator of ONTAP, the world’s
branded storage OS2
in data replication software
#1in serving
Germany
#1
#1
#1Leader in oil &
gas, media &
entertainment,
financial
services,
healthcare
NetApp AFF is #2 with 22.8%
revenue market share#2
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
MongoDB Internal Storage Pain Points
Performance degrades with node and/or drive failures, and increasing users
Difficult to scale up and out on-demand
Time-consuming and costly to manage and maintain
Copying, restoring, protecting and moving data is inefficient and inflexible
4 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Why NetApp
Achieve predictable high performance
Access your MongoDB data more frequently
Scale with ease with compute and storage decoupled
Consume fewer servers and less capacity to lower your cost of ownership
Choose your all-flash certified solution: All Flash FAS, EF-Series, or SolidFire
5 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
High Performance
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only6
NetApp MongoDB Customer Success
7
Large Airline Company
Business Challenge
Utilize analytics to determine new offers
and promotions
Improve performance
NetApp Products: ONTAP AFF
Solution Stack: MongoDB
Why NetApp:
High performance
Good experience with AFF in the past
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
NetApp AFF is MongoDB CertifiedArchitecture
8
Consistent high performance
High throughput/low latency (< 700 μs)
Mongoperf – IO performance tool
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
100% read
Average number of IOPS – 450K
Average Overall Latency – 0.27 ms
50% read and 50%write
Average number of IOPS – 298K
Average Overall Latency – 0.63 ms
AFF MongoDB Performance: Mongoperf
0
0.05
0.1
0.15
0.2
0.25
0.3
0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000 450,000 500,000
Avg.Latency(ms)
#ofIOPS
LatencyGraph100%Reads
100%Reads
Latency and IOPS
9
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0 50,000 100,000 150,000 200,000 250,000 300,000 350,000
Avg.Latency(ms)
#ofIOPS
LatencyGraph50%Reads&50%Writes
50%Reads/50%Writes
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Cost-Effective Performance
10 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
NetApp MongoDB Customer Success Aerospace Company
Business Challenge
Protect IT data assets from cyber attacks
Had difficulty analyzing, processing, and scaling data in a geo-spatial compute cloud
NetApp Solution
FlexPod® Select , NetApp® E-Series, Cisco UCS servers, Cisco Nexus fabric extender
Solution Stack
MongoDB database, Cloudera Hadoop , custom cyber analytics application
Why NetApp
Processing time went from 12 minutes per data cell to 20 seconds with linear scalability
Simple and easy integration—no need to redesign apps or analytic systems
Higher level of reliability for MongoDB and Hadoop
Better TCO vs. commodity servers with internal drives
11 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
NetApp EF is MongoDB Certified
Field proven platform to deliver lowest latency performance
Better performance and reliability vs. commodity servers with internal disks
Performance Read Cache flexibility
12 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Commodity
Servers
NetApp®
EF560
Unavailability
during controller
failure
30 seconds 2 seconds
Performance
drop during disk
failure
92% 20%
Reconstruction
time10 hours 15 minutes
NoSQL: EF560 Proof Points
13
EF560 vs. Commodity Servers with Internal Storage During Failure
EF560 Performance Scaling
0
40000
80000
120000
160000
200000
240000
0 32 64 96 128 160 192 224 256 288 320 352
Th
rou
gh
pu
t(o
ps/s
ec)
YCSB Client Threads (#)
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Enterprise Ready Infrastructure
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only14
NetApp MongoDB Customer Success
Attach Rates 40%
15
Large Bank/Financial Services Company
Business Challenge
Backup/restore for MongoDB
Deployment best practice for Virtual
Infrastructure in NFS and SAN
NetApp Products: ONTAP AFF, SnapCreator
framework, Snapshot, SnapRestore, FlexClone
Solution Stack: MongoDB
Why NetApp:
Demonstrated robust backup and restore
with SnapCreator
Provided deployment guidelines for Virtual
Infrastructure deployment
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Backup & Restore
“User-story” : "I run a mongodump of our databaseevery night. When I did a restore to a totally separatereplica set the sizes are drastically different. I'mgathering that this is probably normal, but looking for alittle more confirmation from the community.
First is the current prod db stats, then the next sectionis the replica set I created and performed amongorestore. Showdbs shows the db at about 140GB,yet the mongodump size is about 80GB.”
Snapshots
16
Source: https://groups.google.com/forum/#!topic/mongodb-user/8MF4Tku8w nI
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Backup & Restore (Cont.)Snapshots
17
Source: https://groups.google.com/forum/#!topic/mongodb-user/8MF4Tku8w nI
Reason: Mongorestore didn’t restore all the collections.
Destination ran out of space or other reasons
NetApp snapshot for backups:
In the case of a restore you get everything you want
the same way as it was when the snapshot was taken
It’s much faster than using mongodump/mongorestore
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
MongoDB: Backup and Disaster Recovery with NetApp SnapMirror
NetApp
AFF
vSphere / vCenter / NetApp VSC
EBS
SecondarySnapMirror
Backup/DR to the Secondary
Fibre Channel
Primary
SnapCreator for
MongoDB
Snapshot backups & replication
Instant database cloning
(test/dev)
NetApp FAS
or
ONTAP
Select
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Eliminate Sprawl of Data Copies
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only19
NetApp MongoDB Customer Success
Attach Rates 40%
20
Photo Sharing Company
Business Challenge
Constant shuffling of data due to
inefficiencies of storage silos
Reduce complexity, and eliminate scaling
challenges
NetApp Products: ONTAP AFF, StorageGRID
Webscale for modified and proxy images
Solution Stack: MongoDB
Why NetApp:
Stability, reliable support
Ease of scalability
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Sharding a CollectionFlexClone
21
Source: https://groups.google.com/forum/#!topic/mongodb-user/D7yw ikq8q4U
“User-story” : "Hello All, I am designing the sharding strategy for one of my collection. Each object in that collection has a timestamp field which I find best to use as my shard key.”
“Forum-comments” : “Shard key selection is an importantstep in your schema design, since once created, the shardkey is immutable. That is, if later on you discover that theshard key is not the best, you would have to dump thecollection, recreate the collection with a different shard key,and re-import all the data back in.”
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Sharding a CollectionFlexClone
22
Source: https://groups.google.com/forum/#!topic/mongodb-user/D7yw ikq8q4U
PRD
mdb server 1
mdb server 2
mdb server 3
NetApp Storage
mdb server 4
mdb server 5
mdb server 6
Cloned PRD
Running shard
key testing
NetApp FlexClone technology
Instant replication without additional storage
space
Easily create an environment to run a shard
key selection testing for your collection
No impact to the production environment (PRD)
No need to re-do everything on the PRD
database, you can promote the cloned PRD to
be the production database.
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
AFF MongoDB Data Reduction with Inline Deduplication and Inline Compression
Test configuration
Test Tool: MongoPerf
Compression disabled on MongoDB side
Result: 15:1 data reduction; saves CPU server cycles for other tasks
NetApp portfolio of storage efficiency technologies helps ensure low storage cost
23
Inline Adaptive Compression
Inline Deduplication Inline Data Compaction
Inline Zero-block Deduplication
All-zero blocks are detected
No user data is written to SSDs; only metadata is updated
• Compresses 8KB logical I/Os into 4KB physical blocks
• 8KB group size aligned with the I/O size used by most databases
Deduplicates incoming blocks against recently written blocks
Use with background (post-write) dedupe to maximize space savings
Combines 2 or more small logical I/Os (typically <2KB) into one 4KB physical block
Also compresses 4KB logical I/Os
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Best of Breed Data Life Cycle Management
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only24
AFF MongoDB on NetApp Data Fabric AFF and Cloud ONTAP in Amazon Web Services
AFF/FAS
EBS
NPS
SnapMirror
Backup/DR to the NPS/SecondaryOn Premises
VMware vSphere 6.0
VMware vCenter NetApp VSC
RHEL RHEL
MongoDB 3.0
FC/iSCSI
RHEL
NFS/iSCSI
AFF/FAS
25 © 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
26
Unified Data Lake for all workloads
Elastic in scale and performance
Spans enterprise, cloud and hybrid
HDFS
Spark
NOSQL NAS
MapReduce Impala
RDBMS
Streaming
Collab
NOSQL
Data Sources
Analytics Engines
Data Lifecycle Management
Backup | DR | Tiering
Datastore Formats
ONTAP Data Lake
(Flash | Disk | Cloud)
NetApp Data Lake Solution
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
NetApp Volume Encryption (NVE)
Secure your data and simplify key management on any type of drive
Zero-management encryption solution for data on disk
Leverage storage efficiency features and encryption; saves CPU server cycles
Volume Encryption and Compliance
SnapLock for Compliance
Comply with strict data-retention regulations and internal IT governance rules
SnapLock Enterprise Administrator is trusted to perform deletes
SnapLock Compliance
Administrator is not trusted to perform deletes
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only27
Inline dedupe and compression reduces MongoDB storage consumption without impacting performance
Integrated API facilitates auto-scaling and full-stack operational integration
Scale MongoDB infrastructure without downtime
Rapid volume cloning capabilities allow for new backup and deployment strategies
MongoDB on NetApp SolidFireSolution overview
28© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Summary
29
NetApp storage solutions maximize your MongoDB Investment:
Accelerate MongoDB performance
Increase availability
Eliminate data sprawl and reduce copies
Streamline data lifecycle management
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Related Sessions and Resources:
30
For additional information after the event visit: http://www.netapp.com/us/solutions/big-data/nosql/mongodb.aspx
MongoDB Solution brief w/ AFF and EF : http://www.netapp.com/us/media/ds-3778.pdf
MongoDB TR w/ AFF: http://www.netapp.com/us/media/tr-4492.pdf
MongoDB TR w/ E-Series: http://www.netapp.com/us/media/tr-4501.pdf
MongoDB Solution brief w/ SolidFire: http://info.solidfire.com/rs/solidfire/images/SolidFire_SolutionBrief_MongoDB.pdf
Visit us at the NetApp booth!
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only
Thank You
© 2016 NetApp, Inc. All rights reserved. NetApp Confidential – Limited Use Only