Analyst Perspective: Storage Capacity to Store Data; Storage

14
Analyst Perspective: Storage Capacity to Store Data; Storage Performance Data; Storage Performance to Deliver Services Benjamin S. Woo Program Vice President, WW Storage Systems Research IDC

Transcript of Analyst Perspective: Storage Capacity to Store Data; Storage

Page 1: Analyst Perspective: Storage Capacity to Store Data; Storage

Analyst Perspective:Storage Capacity to Store

Data; Storage PerformanceData; Storage Performance to Deliver Services

Benjamin S. WooProgram Vice President, WW Storage Systems Researchg g y

IDC

Page 2: Analyst Perspective: Storage Capacity to Store Data; Storage

Th P iThe Premise

• As the way storage purchases are made continues to change, one thing is clear –g , gthe cost per raw gigabyte of storage capacity has become a “poor metric” when purchasing performance-based storage systems,

Page 3: Analyst Perspective: Storage Capacity to Store Data; Storage

C it I thi ?Capacity – Is this new?

202020202009 

0.8 ZB*

202035 ZB*202035 ZB*

*Zettabyte = 1 trillion gigabytes

Source: IDC Digital Universe Study, sponsored by EMC, May 2010

Page 4: Analyst Perspective: Storage Capacity to Store Data; Storage

P f I thi ?Performance – Is this new?

Performance

Time

Page 5: Analyst Perspective: Storage Capacity to Store Data; Storage

Combining Capacity with g p yPerformance

CapacityPerformance

Time

Page 6: Analyst Perspective: Storage Capacity to Store Data; Storage

Reality of capacity to y p yperformance ratio

Capacity

Performance

Page 7: Analyst Perspective: Storage Capacity to Store Data; Storage

C ti b lCreating a balance

$/IOP $/GB$/IOP measures

performance

$/GB measures capacity

Using capacity

optimized di l d

Using performance

optimized di l dmedia leads

to over provisioning of TB’s

media leads to over

provisioning of IO’s

very expensive

very expensive

Page 8: Analyst Perspective: Storage Capacity to Store Data; Storage

Th S l tiThe Solution

• Buy what you need, pay for

• System A– 20TB, costs $60k, p y

what you get– $/GB is inaccurate

20TB, costs $60k• $3k/TB

• System B– 30TB, costs $60k

What about •IOPS•Throughput

• $2k/TB•Throughput•Reliability•P&C•Manageability•Scalability

Page 9: Analyst Perspective: Storage Capacity to Store Data; Storage

Th R litThe Reality

• System A– Included SSD

• System B– FC onlyIncluded SSD FC only

System A has a lower $/IOP

Page 10: Analyst Perspective: Storage Capacity to Store Data; Storage

Wh t di h ld ?What media should you use?

Media Type PerformanceSSD’s >10,000 IOPS15k RPM HDD ~170 IOPS10k RPM HDD ~120 IOPS7200 RPM HDD ~80 IOPS5400 RPM HDD ~50 IOPSCloud ~25 “IOPS” (T1 line)

Page 11: Analyst Perspective: Storage Capacity to Store Data; Storage

Cl d?Cloud?

• Yes, cloud!• Leverage cloud for data that needs to beLeverage cloud for data that needs to be

distributed• Leverage cloud providers that provideLeverage cloud providers that provide

dedupe to lower usage costs better $/GB/month opex

Page 12: Analyst Perspective: Storage Capacity to Store Data; Storage

E ti l G idEssential Guidance

• “Hot” capacity is roughly 2-5% of total storageg– Use SSD’s

• The other 95%?– Use low cost, high capacity drives

• Leverage tiering softwareg g– Make sure it’s automated!!

Page 13: Analyst Perspective: Storage Capacity to Store Data; Storage

Q ti ? Sh t ilQuestions? Shoot me an email

• Ben Woo– VP, Storage

C– IDC– [email protected]

Page 14: Analyst Perspective: Storage Capacity to Store Data; Storage