flash 1st - the storage strategy for the next decade ( emc world 2012 )

1 © Copyright 2012 EMC Corporation. All rights reserved.

FLASH 1st The Storage Strategy for the Next Decade

Denis Vilfort Sr. Director, USD Marketing


2000 2003 2006 2009 2012 2015 2018 2021

140,000

60,000

100,000

40,000

120,000

80,000

20,000

0

Information Tipping Point Ahead The future will be nothing like the past

Data

Budget

Automation

We are

Here

110X MORE DATA

EACH DECADE

A 10TB data center in 2001growing at 60% YoY will be a 120 Petabyte data center by 2021, - but IT budgets remain flat


4,123 TB

Most Data Will be Stone Cold While we increasingly can’t erase data, we can store it better

2011

2021

37.5 TB

Hot Warm Cold

Evolution of 50 TBs in 10 Years

110 X


Why Disk Aggregation is Losing Steam Moore’s Law drives the escalating need for IO transactions

23 36 58 93 148 237 378 603 962 1,536 2,451 3,912 6,243 9,964 15,903 25,381

40,509

64,652

103,184

164,682

262,833

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

The Number of Drives Needed per Host over Time

?

We are

Here


The CPU to HDD Performance Gap CPU improves 100 times every decade – disk speed hasn’t

100 times improved

10,000 times improved FLASH

2000 2010 2020

MOORE’S LAW:

100X per

DECADE

CPU continue to improve while disk drive performance remains flat. As a result, applications will suffer more and more unless we rapidly move to FLASH.


Anatomy of a Enterprise FLASH Drive Designed for reliability, data integrity and performance

SLC NAND FLASH

DRAM

End to End CRC

Controller

SAS or SATA ports


Comparing Associated Costs Which technology is the most efficient?

$- $5 $10 $15 $20 $25 $30

15K HDD

7200 HDD

FLASH

15K HDD 7200 HDD FLASH $/GB $1.80 $0.46 $24.75

Capacity Acquisition Cost

$- $5 $10 $15 $20

15K HDD

7200 HDD

FLASH

15K HDD 7200 HDD FLASH $/IOPS $6.00 $15.28 $0.99

Transaction Acquisition Cost

0 5 10 15 20 25 30

15K HDD

7200 HDD

FLASH

15K HDD 7200 HDD FLASH mWatt/GB 28 4 25

Capacity Power Cost

0.0 50.0 100.0 150.0

15K HDD

7200 HDD

FLASH

15K HDD 7200 HDD FLASH mWatt/IOPS 94.4 133.3 1.0

Transaction Power Cost

Lowest Transaction Cost

Lowest Capacity Cost


Transaction and Capacity support needed by the business? Two Major Components of Your SLA

Needed Capacity

Nee

ded

Tran

sact

ions

$

Lowest $/IO

SSD

Lowest $/GB

NL-HDD

SLA

SLA Target


FLASH Becomes More Affordable Projected to fall below $1/GB by 2018

40% LOWER EACH YEAR

Consumer electronics keep driving FLASH FAB capacity up. As a result, FLASH NAND prices will continue to fall


ROI Improvement Decade Outlook FLASH technology will improve faster than mechanical drives

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

IOPS per $ 0.1 0.3 0.7 1.6 3.7 8.6 20.2 47.1 109.8 256.3 598.0

GB per $ 0.7 1.1 1.9 3.1 5.1 8.6 14.3 23.8 39.7 66.2 110.3

0.1

1.0

10.0

100.0

1000.0

FLASH STORAGE

MECH. STORAGE 29X

MORE IMPROVED


The VNX5500-F High performance, high availability FLASH Array


The EMC VNX 5500-F Flash Array Proven high availability and consistent high performance

• 5 x 9’s availability for mission critical applications

– RAID data protection, proactive global sparing, replication, EMC quality Enterprise Flash

• Starter configuration at 2 or 4TB of Flash

– 25 X 2.5” drives, with 100GB or 200GB SLC SSDs

• Advanced data efficiency services – Compression, De-dupe, and Thin

Provisioning doubles usable capacity

• Full unified protocol support – CIFS, NFS, pNFS, iSCSI, FCP and FCoE

• Expandable to Tiered Storage with FAST – From 49 TB of Flash to 675 TB of tiered

storage (Flash, SAS, NL-SAS) as data ages

10X TRANSACTIONS

1/8th THE COST


The FLASH 1st Data Management Strategy Hot data on fast FLASH SSDs—cold data on dense disks

“Hot” high activity As data ages, activity falls,

triggering automatic movement to high

capacity disk drives for lowest cost

Highly active data is stored on

FLASH SSDs for fastest

response time

Movement Trigger

FLASH SSD

“Cold” low activity

Dat

a Ac

tivi

ty

High Cap. HDD

Data Age (5 years)


Gaining Economic Efficiency FLASH SSDs for active transactions & 3TB HDD for overflow

• FLASH SSDs offers highest

transaction efficiency

• Large capacity HDD offers highest capacity efficiency

• High RPM HDDs are not capacity-efficient nor are they performance-efficient

• FAST VP automatically and seamlessly optimizes data across all disk technologies

0

0.2

0.4

0.6

0.8

1

1.2

0 0.5 1 1.5 2 2.5

IOPS

per

$

GB per $

FLASH SSD

15K HDD 3TB HDD

FAST VP


Today’s Compute Hierarchy Right data. Right Place. Right Cost.

Multi-Core/Socket CPUs − pS latency

DDR4 - 4.266GHz RAM − 7 to 200nS latency

200GB FLASH SSDs − 20 to 320 uS latency

3TB HDDs − 7 to 34 mS latency

pS

nS

uS

mS

FAST Cache

FAST Virtual Pools


Lowest $/IOPS and lowest $/GB Optimizing Your VNX Deployment

Max System IOPS

System Scale

Tran

sact

ions

Add SSDs for more IOPS

Add HDDs for more Capacity


Hybrid Systems A little FLASH goes a long way

30 75 180

415

910

45 50

70

85

90

VNX5100 VNX5300 VNX5500 VNX5700 VNX7500

SSDs HDDs


The FAST Suite Dynamically optimizes IO for FLASH 1st at a 64K page size

• FAST Cache – Caches data from the

HDD or NL-HDD tiers in the pool

– Operates at a page granularity of 64K

• FAST VP – Dynamically moves data

between tiers in the storage pool

– Operates at a slice granularity of 1GB

• Deploying both together ensures maximum IO granularity

64K 64K

64K

Controller

FAST Virtual Pool

FAST Cache

NL-HDD HDD SSD 1G

FAST Cache - FAST VP Relationship


IO Skew Random slice activity

• Slice activity stack ranked according to access frequency of a given slice

• 51 slices of 101 has 80% of IO

0

100

200

300

400

500

600

700

800

900

1000

1 7 13

19

25

31

37

43

49

55

61

67

73

79

85

91

97

All

80%

0

100

200

300

400

500

600

700

800

900

1000

1 6 11

16

21

26

31

36

41

46

51

56

61

66

71

76

81

86

91

96

101

• Slice activity 100% randomized


IO Skew Random slice activity weighted by 3% cooling rate

• Stack ranked slice map weighted by a 3% cooling rate

• Now only 31 slices of 101 total has 80% of IO

0

100

200

300

400

500

600

700

800

900

1000

1 6 11

16

21

26

31

36

41

46

51

56

61

66

71

76

81

86

91

96

101

All

"80%"


Examples of IO Skew Driven by Data Growth and Business Models

Very High Skew

Growth: 60% Days Hot: 10 1% of data = 80% of IO

High Skew

Growth: 100% Days Hot: 30 4.2% of data = 80% of IO

Low Skew


Std. Skew



How Much FLASH? Used dynamically with FAST

• Data follows a predictable decay in activity

• Older data is constantly being replaced by new highly active data

• The amount of FLASH required is determined by:

– The amount of data created each day, and

– The period of time it takes to cool

Amount of Data being created daily

Number of Days of High Data Activity

FLASH Capacity

Size of FLASH FIFO


Modeling “Data Decay” A simple GB-Day state model describes typical data behavior

Prob

abili

ty o

f dat

a ac

cess

1GB data age in days

P = (1-C) A, and A = LOG1-C (P)

Where: P is Probability of data access C is daily Cooling rate A is data Age in days

Hot


Modeling “Data Decay” A simple GB-Day state model describes typical data behavior

1GB data age in days

Hot Cold

Hot

Co

ld

State Change

P = (1-C) A, and A = LOG1-C (P)

Where: P is Probability of data access C is daily Cooling rate A is data Age in days


Basic Data Decay Model FL

AS

H T

ier

FAS

T

Cac

he

15

K

7 days 52days

20%

60%

1%

250 days

• The basic storage pool distribution model assumes FLASH as the first line of defense for 80% of all IOs

• As an additional performance buffer, 15K drives are used for overflow for 19% of all IOs

• Only a 1% chance remains that any IOs will be serviced from slow 2TB drives


How Long Does Data Stay Hot? Depends on business model, applications and workload

Days of high data activity 30 days 60 days 90 days 120 days

Hot Days Daily

cooling rate @ 80%

30 5.2%

60 2.7%

90 1.8%

120 1.4%

Hot

Co

ld


FLASH Percentage A function of data growth and service level

• Relative more FLASH is needed when:

– Service level (FLASH hit rate) is elected high, and

– Data growth is high

• More than 25% FLASH is highly unlikely!

Data Growth Rate

FLA

SH

por

tion


Getting the Right Blend of Flash 3 fundamental business questions

• How much data is under management today?

30 TBs

• How much is your data growing each year?

50% YoY

• How long does your data stay hot?

60 days


Calculating Net New Data Net new is a function of amount of starting data and growth

• 50% of 30TB is 15TB

• The average amount of data generated each day:

– 15 X 1024 GB / 365 = 42 GB per day

50% Growth Rate

30 30

Start Year 1

15


Calculating Needed FLASH FLASH only needs to be large enough to hold hot data

• FLASH capacity: – 60 days X 42 GB =

2,520 GB

• FLASH Percentage: – FLASH Capacity/Total

Capacity – 2,520 GB/(45 X 1024) GB X

100 = 5.4%

30 30

12.5

2.5

Start Year 1

Needed FLASH

FLASH portion


Calculating How Much FLASH

FLASH Portion

FLASH % = Yearly Growth Rate% X Number of Hot Days X 100

365 X (Yearly Growth Rate% + 100%)

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

30 days Hot 1% 1% 2% 2% 3% 3% 3% 4% 4% 4%

60 days Hot 1% 3% 4% 5% 5% 6% 7% 7% 8% 8%

90 days Hot 2% 4% 6% 7% 8% 9% 10% 11% 12% 12%

120 days Hot 3% 5% 8% 9% 11% 12% 14% 15% 16% 16%

0% 2% 4% 6% 8%

10% 12% 14% 16% 18%

FLA

SH

por

tion

o

f sto

rage

poo

l

FLASH portion as a function of Yearly Data Growth


Using the Business Data to Configure With no trace data, the model is used

SSD

NL-HDD

Hot

Co

ld

60 days

178 days

15K-HDD

3-Tier FLASH 1st Strategy

• The SSD tier retains a 80% probability of data re-use

• An 15K HDD tier is used as overflow with a 19% chance of data re-use

• Finally, the NL-HDD tier is used for long term online storage with a 1% chance and lower of data re-use


Smaller Foot Print SSDs for FAST Cache 4 SSDs for FAST VP 20 15K SAS HDDs for FAST VP 15 NL-SAS HDDs for FAST VP 24 TOTAL 63

$182,882

VNX5300

Configuration Example 30TB usable , 50% growth, 60 days hot @ 80% FLASH service

Monolithic

Large Foot Print 200 SAS 15K HDDs

VNX5500

$380,573


Tangible Business Benefits Dramatically lower cost and higher transaction support

$8.26

$3.97

Without FAST

With FAST

$ per Usable GB

200

63

Without FAST

With FAST

Disk Slots Needed

30,660

7,030

Without FAST

With FAST

Yearly Power Consumption

15,050

36,686

Without FAST

With FAST

Transactions Supported


12.5X Compounded Improvement in ROI $/GB and $/IO is improves simultaneously

0.1 GB/$ 0.04 IO/$

0.25 GB/$ 0.20 IO/$

0.00

0.05

0.10

0.15

0.20

0.25

0.00 0.05 0.10 0.15 0.20 0.25 0.30

IOPS

per

Bud

get $

GB per Budget $

ROI Improvement from FLASH 1st

5X

2.5X


Data Strategy Matrix Technology differentiated data placement strategy

Low Activity High Activity

Wri

te

Read

Dedicated PCIe FLASH Shared NL HDD

Shared FLASH SSD Shared 15K HDD

Automation


32 cores 1M IOPS

32 cores 1M IOPS

32 cores 1M IOPS

32 cores 1M IOPS

Virtualization Further Pressures Storage Moore’s law and better CPU utilization drives IOPS need

NEED for SPEED

• Application management - like vMotion – requires shared storage

• A performance budget of 1M IOPS per host is now a reality

• Shared storage risks becoming bottleneck in high performance virtualization deployments

200K IOPS

2M IOPS

Note: 2M host pressure IOPS is based on 50% storage side IOPS per host


VNX Virtualized Cloud Architecture Host-side read caching multiplies performance

• Scale out for “hot data” and read performance with FLASH based host-side caching

• Scale up for deep archive of “cold data”

• All writes are protected by synchronous cache mirroring to DRAM & FLASH in central array

Performance Scale

Cap

acit

y S

cale

10X BETTER

PERFORMANCE

2M IOPS

Note: 2M storage IOPS is based on 90%/10% read/write mix


High Performance Storage Hierarchy The data continuum is FLASH centric and extends to the host

• Host-side read FLASH caching shields data from SAN and array latency

• Array-side FLASH protects data against host failure

• Array-side HDDs accumulates inactive data at lowest cost

HDD

SSD FLASH

PCIe FLASH

50uS

200uS

1,500uS W

orking Set

Inactive D

ata


The IO Spread: Access Density PCIe FLASH offers 8,333X better data access than 15K HDD

0.03 0.3

25

2,500

0.01

0.1

1

10

100

1000

10000

NL-HDD HDD SSD PCIe FLASH

Access Density: IOPS/GB


Data Continuum As data activity falls, different technologies apply

“Hot” high activity

“Cold” low activity

Dat

a Ac

tivi

ty

Data Age

VFCache

SSD FAST Cache

SSD FAST VP

15K HDD

NL-HDD

OPTIMIZED COST Dynamic Storage

Hierarchy differentiated on Access Density Cost


USD Lectures- Session Name Day & Time

#7 VNX Storage Efficiencies – What, Why, and When Monday 10:00am - 11:00am Tuesday 11:30am - 12:30pm

#8 VNX Virtual Provisioning - Leveraging the Best in Efficiency and Simplicity with Virtual Pool Provisioning

Tuesday 10:00am - 11:00am Wednesday 11:30am - 12:30pm

#9 VNX Compression and Deduplication - Optimizing Capacity for Your File and Block Infrastructure Tuesday 4:15pm - 5:15pm Thursday 10:00am - 11:00am

#10 VNX FAST VP - Optimizing Performance and Utilization of Virtual Pools Tuesday 11:30am - 12:30pm Thursday 8:30am - 9:30am

#11 VNX FAST Cache – Super Charge your Storage with Extended Cache Monday 4:00pm - 5:00pm Wednesday 10:00am - 11:00am

#12 FAST Storage Design Basics for EMC VNX Tuesday 2:45pm - 3:45pm Wednesday 4:15 - 5:15pm

#13 Leveraging SSD: Designing for FAST Cache and FAST VP on Unified Block Storage Wednesday 11:30am - 12:30pm Thursday 11;30am - 12:30pm

#25 FLASH 1st - The Storage Strategy for the Next Decade Monday 8:30am - 9:30am Monday 2:30pm - 3:30pm

FBU#1 Introduction To EMC VFCache Monday 8:30am - 9:30am Wednesday 4:15pm - 5:15pm

FBU#2 Leveraging EMC VFCache with Enterprise Applications Monday 4:00pm – 5:00pm Thursday 10:00am - 11:00am

BOF Dive into EMC VNX & FAST Suite Storage Efficiencies Tuesday 1:30pm - 2:30pm

BOF Flash Technology in the Storage Environment Wednesday 1:30pm - 2:30pm

Storage Efficiencies/Flash Sessions


Provide Feedback & Win!

• 125 attendees will receive $100 iTunes gift cards. To enter the raffle, simply complete:

– 5 sessions surveys – The conference survey

• Download the EMC World Conference App to learn more: emcworld.com/app


FLASH 1st

THANK YOU

Denis Vilfort Sr. Director, USD Marketing