12
1 | © 2015 Pure Storage Inc. Why Pure Storage for MongoDB CRAIG HALLIWELL STRATEGIC ALLIANCES, PURE STORAGE JUNE 1, 2015 NOSQL ON PURE STORAGE

Consolidate and Simplify MongoDB Infrastructure with All-flash

  • Upload
    mongodb

  • View
    122

  • Download
    0

Embed Size (px)

Citation preview

1 | © 2015 Pure Storage Inc.

Why Pure Storage for MongoDB

CRAIG HALLIWELLSTRATEGIC ALLIANCES, PURE STORAGE

JUNE 1, 2015

NOSQL ON PURE STORAGE

2 | © 2015 Pure Storage Inc.

Agenda

• Shared Vision

• MongoDB 2.x on Pure Storage

• Changes with MongoDB 3.0

3 | © 2015 Pure Storage Inc.

Shared Vision – Change the “Status quo”MongoDB’s goal is to make it easy to code, scale and manage databases

Pure’s goal is to make it easy to provision, scale and manage storage

For every 250 TB of Tier 1 storage*,

Pure Storage can free up:

1/2 Employee

+

$500K per year

4 | © 2015 Pure Storage Inc.

MongoDB 2.x Data Reduction on Pure Storage

RITA collects all this flight information, archives it and distributes it

5 | © 2015 Pure Storage Inc.

Data Reduction on RITA Database [MongoDB 2.6]14 years of flight data, ~500GB MongoDB database size, Only 47GB on Pure

6 | © 2015 Pure Storage Inc.

BUSINESS & IT TRANSFORMATION – Case Studies

• Data Reduction: 3:1 to 15:1 (4.5:1 Average) /

• Production Platform for MongoDB (75TB Instance)

• Massive Operational Cost Reduction: Decreased OpEx Costs over $750,000 over 3 year by reducing storage datacenter footprint from 9 racks to 2.5 racks at a fraction of the power / cooling requirement.

• User Experience: Response times are down from 4 – 8 ms response time to 1 ms.

• Pure Simplicity: Deployed 750 TB of storage in 7.5 hours with 4 people and no professional services costs. FTEs required to support 1.2 PB = 1 FTE.

• MongoDB 2.x instance is getting 15.8:1 data reduction resulting in $1.82/GB and datacenter reduction from 40RU down to 8RU.

• Consolidated 24TB in all environments to ~4TB on Pure with > 6 to 1 data reduction

• Savings of $130,000 in spinning disk

• Grown 30% since July and now represents 32TB of data occupying ~5TB on Pure

• Performance issues have ceased and we are getting sustained 1ms response times

• Also migrated multiple MySQL, SQL, and critical custom apps freeing up older Tier 1 spinning storage array

• 4.7 to 1 data reduction of applications residing on Pure

• 8U footprint for 80TB of Tier 0 Storage

7 | © 2015 Pure Storage Inc.

MongoDB 3.x Changes Everything!!• MongoDB 3.x Now Available

Significant improvementsin database & IO layers

MongoDB 3.x can compressdata when using WiredTiger

Dramatically simplifiesmanagement

Can Pure Storage still deliver any of its saving with MongoDB 3.x???

8 | © 2015 Pure Storage Inc.

Don’t Let Storage Be Your BottleneckRealize all improvements in MongoDB 3.x database & IO layers with Pure

mmap WiredTiger (snappy)0

1

2

3

4

5

6

1.0

5.3x

Relative transactions•Harvest the performance benefits from WiredTiger IO pluggable engine while enjoying its capacity savings

•WiredTiger IO engine performs 5.3 times better than mmapv1

9 | © 2015 Pure Storage Inc.

Can Pure Storage Compress Any Further? Pure Storage further compresses data from WiredTiger

mmap WiredTiger (snappy) Pure Storage0

50

100

150

200

250

300 282

7237

Space consumed (GB) by database on MongoDB 3.x

2:1 data reductionon top of WiredTiger compression

10

| © 2015 Pure Storage Inc.

Can Replica Sets Improve Data Reduction? Pure Storage improves data reduction slightly further (1 primary & 3 replicas)

WiredTiger (snappy) Pure Storage0

50

100

150

200

250

300

350

400 372

167

Total Space Consumed by database & 3 replicas (GB)

•Replica sets are not guaranteed to be physically identical to each other, impacting deduplication

• Improved data reduction may be possible with bigger documents

•Leveraging Pure Storage snapshots for seeding the replica sets can improve data reduction further

2.2:1 data reductionon top of WiredTiger

11

| © 2015 Pure Storage Inc.

MongoDB Rapid Cloning using Pure Storage Snapshots

MongoDB

Redhat

MongoDB

Redhat

Production Test / Dev

THANK YOU