Vous êtes sur la page 1sur 31

Storage Efficiency &

Optimization Review
Storage Optimization
Maximizing Your Current Storage Assets

A presentation to:
John Doe
XYZ Company
Month XX, 2008
Executive Summary

• Today’s Storage World


• Storage Utilization
• Case Study
• MSI Storage Efficiency & Optimization
Review
• Conclusions
• Q&A
Storage Spend - Major Budget Item
Peripherals
• US annual spend on 9%

storage $25+ billion Networking Servers


27%
• Storage is the 2nd 19%

highest cost category


for US IT departments PCs/
Storage
• Storage needs are Notebooks
22% 23%
growing 50%+ per Source: IDC

year in most large Industry Quotes:

enterprises IDC Perspectives Study, 2008 “Storage is


the 2nd highest percentage of IT spend”

The InfoPro, Wave 11: “Storage spend


growth is over 50% per year in average
Fortune 1000 companies”
How much space is that?
Information object How many bytes Information object How many bytes
A binary decision 1 bit 50,000 trees made into paper and printed 1 terabyte ( TB )
A single text character 1 byte
An academic research library 2 terabytes
A typical text word 10 bytes
The print collections of the U.S. Library 10 terabytes
A typewritten page 2 kilobytes ( KB s) of Congress
A low-resolution photograph 100 kilobytes The National Climactic Data Center 400 terabytes
A short novel 1 megabyte ( MB ) database

A high-resolution photograph 2 megabytes Three years' of EOS data (2001) 1 petabyte ( PB )


The complete works of Shakespeare 5 megabytes All U.S. academic research libraries 2 petabytes
A minute of high-fidelity sound 10 megabytes All hard disk capacity developed in 1995 20 petabytes

One meter of shelved books 100 megabytes


All printed material in the world 200 petabytes
The contents of a CD-ROM 500 megabytes
Total volume of information generated in 2 exabyte s ( EB s)
A pickup truck filled w/books 1 gigabyte GB ) 1999
The contents of a DVD 17 gigabyte s All words ever spoken by human beings 5 exabytes
A collection of the works of Beethoven 20 gigabytes

Tables are derived from figures calculated by Roy Williams on his page called "Powers of Ten." Although his
page has expired, another version derived in part from it is located at the University of California at Berkeley.
Business As Usual?
• It is cheaper to buy new rather than optimize existing
– We’ll dispel this by showing you your own numbers
• Budgets crimped; storage professionals asked to “do
more with less”
– We’ll show you how to squeeze every last bit out of
your existing equipment and show you how to make
the right investments going forward
• Storage is “too complicated”
– It’s hard work, but not too complicated when you have
the right tools and the right partner
How Did We Get Here?
• Everyone has the best interest of the business in mind;
not always the right approach however
• Running out of storage is BAD! Overcome your fears
• The “doubling” factor and other math problems
• Old School: Take all you want – just don’t try to give any
back
• John Madden obvious – “We’re not hoarding; We’re just
taking extra to make sure we don’t run out.”
• “I can buy a terabyte disk for xx dollars at Best Buy…” -
It’s not just the disk!
Storage Utilization
What is your REAL storage utilization?

• Ask three internal sources, get three different answers (Our Case Study
explains why all three can be right and wrong at the same time)

Why?

• Everyone is “missing” data; nature of Storage design


• Lack of inter-department communications
• No tools, ineffective tools, inaccurate tools
• “Special case” data collection problems – VMWare, Microsoft Hyper-V,
Linux Zen, AIX LPAR’s, Solaris Zones, clustering, raw volumes
• Under the restrictions of last decade’s technologies
• Simply not easy to measure and harder to fix utilization issues
• Not anyone’s problem; is *everyone’s* problem
Case Study
XYZ Corporation: Understanding REAL Storage Utilization

• Case Study reflects real results from many


past engagements – not a worst case
scenario
• XYZ Corporation
– A single data center
– A mixture of storage from many
manufacturers
– Experiencing moderate storage growth driven
by new applications, video capture new data
warehousing projects
Case Study
What is your REAL storage utilization?
Case Study: Data Center Storage
The Storage Manager’s View

EMC DMX4 – HP XP24000 – IBM DS8300 –


200TB, 90% full 225TB, 80% full 120TB, 75% full

Allocated Capacity Analysis


Data Center: Total Capacity Available Total Allocated
DMX4 200TB 20TB 180TB
XP24000 225TB 45TB 180TB
DS8000 120TB 30TB 90TB

Totals 545TB 95TB 450TB

Percentages 100% 17% Available 83% Utilization


Case Study: Data Center A
The Storage Manager’s View
• 83% *allocated* utilization – time to buy new storage! Hold on there
partner…
• Allocated vs. Consumed – a huge difference!
• Allocated utilization percentage *is* important; once you hit 100%
full you are out of storage until something changes
• Without a reliable Storage Resource Management (SRM) tool
collecting data from the entire storage environment there are no
other data points to calculate a different utilization rate
• A good SRM vendor will teach client storage team members how to
interpret and analyze SRM reporting
• Agents on servers is the only practical way to collect more detailed
information about storage usage and arrive at the REAL utilization
Case Study: Orphaned Storage
The Storage Manager’s View
• Most basic form of storage reclamation: Orphaned Storage
Reclamation
• Occurs when servers physically move or disconnect from the
storage network and no one notifies the storage team
• 10-15% of a data center’s capacity can be tied up in orphaned
storage at any given moment
• Easy to identify with simple SRM reporting
• Each instance must be investigated to insure the storage is really
orphaned; servers that are powered off can look like they have
orphaned their storage
• Orphaned LUN reclamation needs to be part of standard operations
and performed either weekly, monthly or quarterly
• Best case scenario is server teams notify storage when moving
servers (via work order, service request or other mechanism)
Case Study: Orphaned Storage
The Storage Manager’s View

Allocated Capacity Analysis w/Orphaned Disk Capacity


Data Center: Total Total Available Total Allocated
Capacity
Orphaned Available
DMX4 200TB 20TB 20TB 160TB
XP24000 225TB 22.5TB 45TB 157.5TB
DS8000 120TB 12TB 30TB 78TB

Totals 545TB 54.5TB 95TB 395.5TB

Percentages 100% 10% Available 17% Available 73% Utilization

• Note that Allocated Storage decreases and orphaned disk


becomes available disk
• Utilization drops to 73% from 83%
Case Study: This Is EASY!
Things Get a Little More Complicated…

In BIG
trouble!
Case Study: Storage Utilization
The Server Admin’s View

Oracle Files Layout (9i best practices example)

The following table lists the Oracle files layout for Oracle9i Database Certified Configuration. In the table, SID represents DBNAME for single
instance databases, and DBNAME plus instance number for Real Application Clusters databases.

File System/ Description


Directory
/SID/arch File system for database archive logs.

/SID/backup File system for database backup files. This partition is currently only created for the Production instance.

/SID/oracle File system containing the Oracle database binaries and configuration files
/SID/oradata File system containing data01, data02 and data03 directories.
/SID/oradata/data01 All data files except system and temp data files. For example, data files for Oracle Portal, Oracle Files,
e-mail, tools, and so on. These are mainly the *data*.dbf, *idx*.dbf, and control01.ctl files.

/SID/oradata/data02 Contains the system and temp files. These are mainly the system*.dbf, temp*.dbf, and
control02.ctl files.
/SID/oradata/data03 Log files only. These are mainly the log*.dbf, and control03.ctl files.

Table 1-2 Oracle9i Certified Configuration Oracle FIles Layout


Case Study: Storage Utilization
The Storage Admin’s View

• Storage Request Workflow…


• request given to storage, Storage
Administrator processes request and
assign’s LUN’s to server(s)
Case Study: Storage Utilization
The Server Admin’s View

• LUN’s are used by Server


Oracle Data
Admin to create file systems

Oracle Program Files

Backup

Archive Logs

• Standard practice that


Reserve
some LUN’s held in
“reserve” for future growth
Case Study: Storage Utilization
Reserve LUN’s

• 20-50% of your storage can be tied up on “reserve”


LUN’s
• Convenient for Server Admin, expensive for the
organization
• A practice born from distrust or lack of design – fear
storage won’t be there when needed; fear that
storage team can not execute quickly enough
• “It’s not hoarding, I am just keeping a little extra in
case I need it.”
• Can be fixed with good engineering design, finely
tuned operations and open communications between
storage and server admins
Case Study: Reclaimed Reserve
The Storage Manager’s View
Allocated Capacity Analysis w/Orphaned Disk Capacity + Reserve
• Note that Allocated Storage
decreases and orphaned disk
becomes available disk
• Data Center:
Utilization Capacity
drops to 73% from Total Available Allocated

83% Orphaned Available Reclaimed


Reserve
DMX4 200TB 20TB 20TB 36TB 124TB
XP24000 225TB 22.5TB 45TB 36TB 121.5TB
DS8000 120TB 12TB 30TB 18TB 60TB

Totals 545TB 54.5TB 95TB 90TB 305.5TB

Percentages 100% 10% Available 17% Available 17% Available 56% Utilization
Case Study: File System Utilization
Filesystem Data

• Typical average file system utilization across


an organization is 30%
• Some file systems contain more disk than
needed for performance reasons (more
spindles)
• LUN size is important; a variety of small,
medium and large LUN’s is best
• “Right sizing” is difficult without any history or
accurate data regarding growth
Case Study: Utilization Trending
Utilization Trend
UtilizationTrendingw/Optimization

90

83
80

73
70

60

56

50

40

30

20
1
7

1
0

0
WithoutAnalysis OrphanLUN Reserve Filesystem

AnalysisSteps
Case Study: File System Utilization
Filesystem Data

• Filesystem data optimization opportunities:


– Backups to disk*
• Purging after expiration?
• How long and how many are they kept?
– Duplicate data
– Old data*
– Inflated databases
• Investigate auto growth
– Archival data*
– Log file growth (purging/pruning routines?)*
• How much log data is kept?
– Unused filesystems*
* Potential for secondary tier of disk?
Case Study: Conclusions
Opportunities
• Reclaim orphaned LUNs
• Reclaim reserve LUNs
• Reclaim unused allocations
• Shrink filesystem data
• “Re-platform” servers with “right sized” disk allocations
• Move non-critical data to secondary tier of disk
• Create storage friendly policies and procedures that
prevent over allocation and misuse of disk
• Examine technologies that can help prevent future
issues with disk allocations: thin-provisioning, storage
virtualization, tiered storage
The MSI Offering
Storage Efficiency & Optimization: Stair Step Approach

Data Collection

Analysis

Notification

Remediation

Prevention
The MSI Offering
• Storage Efficiency & Optimization Review engagement
that requires no new hardware or software investment
• Stair step approach
• Goal of letting the customer choose how much
involvement they want
• Data Collection usually completed within several weeks;
Other steps are dependant upon the size of the
environment and the number of optimization
opportunities uncovered in Data Collection; typically four
to eight weeks
• Installation and temporary use of MSI SRM tool as part
of overall offering is required; software removed upon
completion of engagement
• Project management included throughout each step
Policy Changes
• Must be top down; lateral policy implementation is
ineffective
• Must be a common understanding between departments
why a policy is in place; the policy is for the common
good of the company
• Must provide alternative means of accomplishing
business goals if a new policy negates long standing
habits/processes
• Must be enforceable and monitored
• When appropriate, consider historical data in formulating
or modifying policies
• Continue to review and fine tune policies over time; is not
a “one and done” activity
What is the Opportunity?
• Simple math; if 80% allocated can be reduced to
20% allocated the environment can double twice
without any new disk purchases
• The three year total cost of one PB of Enterprise
Storage is $10 million*; shrinking from 80%
allocated on one PB to 20% allocated allows for
4x growth and avoids growth to 4PB – net
savings or 3 x 15 million equals $30 million
cost savings over three years
* Conservative estimate based on standard industry pricing. Please use your own cost numbers to
calculate your own savings – most likely the savings potential will be HIGHER!
Total Cost of Storage
• “All in” three year cost of storage is $10-$20/GB of disk purchased and includes:
– Storage frame and disk
– Management software
– Replication and copy software
– Fiber switching
– Fiber cabling
– Management consoles
– Installation labor
– Operational Support labor (storage, help desk, network, etc.)
– Engineering and Architecture labor
– Management labor
– Heating and cooling
– Data center footprint
– Electricity
– Maintenance and vendor support
– Parts and components
– Decommissioning (labor, scrubbing & transportation)
Storage Optimization
Cost Savings Examples

Cost Savings Analysis


Current REAL Overall REAL Utilization New REAL New REAL 3 Year $ Savings
Utilization % Capacity in In GB's Utilization % Utilization In
GB GB's
10% 1,000 100 20% 200 $ 10,000,000
10% 1,000 100 30% 300 $ 20,000,000
10% 1,000 100 40% 400 $ 30,000,000
10% 1,000 100 50% 500 $ 40,000,000
10% 1,000 100 60% 600 $ 50,000,000
10% 1,000 100 70% 700 $ 60,000,000
15% 1,000 150 30% 300 $ 10,000,000
20% 1,000 200 80% 800 $ 30,000,000
25% 1,000 250 50% 500 $ 10,000,000
30% 1,000 300 60% 600 $ 10,000,000
35% 1,000 350 70% 700 $ 10,000,000
* Assuming three year all in cost of enterprise disk is $10/GB. Example in red.
Conclusions
• MSI engagement can potentially save companies
millions on storage purchases; Potential for HUGE
payback
• Will highlight what storage team is doing right and help
remediate what needs additional work
• Helps in implementing policies that can prevent future
problems/issues
• Because MSI has expertise and experience in these
engagements they can be accomplished quickly and
quietly in an organization; potential for some “quick
wins”
Thank you!

Questions?

Vous aimerez peut-être aussi