SlideShare a Scribd company logo
Storage for
the DBA
Joey D’Antoni
February 13, 2018
Joey D’Antoni
Joey has over 20 years of experience with a wide
variety of data platforms, in both Fortune 50
companies as well as smaller organizations
He is a frequent speaker on database administration,
big data, and career management
MVP, MCSE BI and Data Platform
VMWare vExpert
He is the co-president of the Philadelphia SQL Server
User’s Group
He wants you to make sure you can restore your data
Agenda
Storage Terminology
Array Cache Setup
RAID Types
Tiered Storage
Disk Alignment
Spindle Types
Physical Array Diagram
Storage Terminology
LUN = Logical Unit Number
Host = The Server or Servers a LUN is presented to
SAN = Storage Area Network
Fabric = Fibre network which makes up the SAN
Array = Box with the Spindles in it
Array Diagram (Legacy)
Block and Object Storage
Storage Terminology
Disk = How the OS sees a LUN when presented
Spindle = Physical disks in the Storage Array
IOps = Physical Operation To Disk
Sequential IO = Reads or writes which are sequential on
the spindle
Random IO = Reads or writes which are located at
random positions on the spindle
Array
Cache
Setup
OLTP databases make
poor use of SAN read
cache
OLAP databases make good
use of SAN read cache
Try reducing read cache and
increasing write cache
OLTP databases with high
buffer cache hit ratios may be
able to have the read cache
disabled
There is no one correct
setup. Every system is
different.
When Write Cache Gets Full?
Doesn’t flush to
disk until low
watermark is hit
Force flushes
once high
watermark is hit
Force flushing
completely
empties write
cache
Force flushing
disables write
cache until write
cache is disabled
If adjustable, set
low very low, and
high watermark
very high (20/90)
Oh No, the power went out!
Most arrays
have internal
batteries
Write cache
is flushed to
disk
After flush
array powers
down
On power-up
flushed
cached is
read and
committed to
LUNs before
LUN is made
available
Make sure
SQL is down
before the
array flushes
http://www.flickr.com/photos/robertfrancis/352039299/
Evolution of Arrays
Spinning
Disk
Hybrid All SSD
Spinning Disk and Hybrid Arrays
RAID 0
Straight Stripe
No redundancy
Very fast
Low cost
Any disk failure looses data
Can not survive no disks failures
Requires 2 or more disks
RAID 1
Full Mirror of data
No performance Benefit
High Cost
Requires 2 disks
Can survive 1 disk failure
RAID 0+1
Drives Striped, then Mirrored
High Cost
High Performance
Requires 2 disks
Can survive 1 spindle failure
Requires even number of disks
May survive multiple spindle failure
RAID 10 (1+0)
Drives Mirrored, then Striped
High Cost
High Performance
Requires 2 disks
Can survive 1 spindle failure
Requires even number of disks
May survive multiple spindle failure
RAID 5
Low Cost
Requires 3+ disks
Stripe with Single Parity
Can survive a single drive failure
Write Penalty
Good Performance
RAID 6
Stripe with 2 parity
Higher cost per gig than RAID 5
Requires 4+ disks
Can survive 2 disk failures
Similar penalty as RAID 5
Not all RAID 5 is Created Equal
RAID 6 can be done this way as well.
Disk Alignment
An old problem, that can still be an issue
By default old Windows disks are offset incorrectly giving a 50% performance penalty
Correcting this can improve SQL disk performance up to 100%
• (64 1k blocks/64k IO)=100% of IO is impacted
Must be done before data is put on the disk
Windows 2000 - Diskpar
Windows 2003 - Diskpart
Windows 2008+ - Automatic
Disk Alignment
Auto-Tiering Storage
Used in hybrid
arrays
Gives best
performance
storage for some
types of data
Puts hot data
SSD
Puts idle data on
spinning disks
SSDs Are Fast, But Different
Random I/O becomes a good thing
Compressing data can be done by the array, very efficiently
All flash SSDs tend to have a different platform architecture
These arrays are disruptive to the market in ways beyond
pure I/O performance
NVMe and NVDimm
Thin Provisioning
Storage in the Cloud
Ten Tips to Healthy Storage
Be honest with your storage team
Give back storage when you don’t need it, or know
you won’t need it for a while
If there’s a performance problem, make sure it isn’t
your database first
It helps if your storage admin understands databases
SSDs require different configuration—understand
your array
Ten Tips to Healthy Storage
Always compress your backups (you need to fight this one)
Advanced technologies like NMDimm are good for high-end
workloads
More memory is always better
You can’t overcome terrible storage performance
The cloud is different, and you may be more in charge

More Related Content

PPT
SSDs - Improving Performance of Storage Arrays
PPTX
SSD - Solid State Drive PPT by Atishay Jain
PPT
Hardware2_Charity
PPT
SSD PPT BY SAURABH
DOCX
Raid- Redundant Array of Inexpensive Disks
PPTX
Solid state drives
PPTX
Solid State Drives (Third Generation) 2013
PDF
SSD vs HDD - A Shift In Data Storage by Todd Dinkelman
SSDs - Improving Performance of Storage Arrays
SSD - Solid State Drive PPT by Atishay Jain
Hardware2_Charity
SSD PPT BY SAURABH
Raid- Redundant Array of Inexpensive Disks
Solid state drives
Solid State Drives (Third Generation) 2013
SSD vs HDD - A Shift In Data Storage by Todd Dinkelman

What's hot (20)

PPT
Solid State Drive (SSD) - SBMathema
PPTX
Raid
PPTX
Ssd – Solid State Drive
PDF
ssd vs hdd infographic
PPT
Raid : Redundant Array of Inexpensive Disks
PPTX
What is R.A.I.D?
PDF
Introduction to Solid State Drives
DOCX
Solid State Drives - Seminar Report for Semester 6 Computer Engineering - VIT...
PPTX
SC12 S3700 Series launch event gold
PPTX
Ssd(solid state drive )
PPTX
Solid state drive (ssd)
PPTX
Solid state drive
PDF
Trio nas qsan storage
PDF
Enterprise SSD: Who is Adopting Them and Why
PPTX
Understanding Solid State Disk and the Oracle Database Flash Cache (older ver...
PPT
Solid state solid state drives
PDF
Solid State Drive Technology - MIT Lincoln Labs
PDF
Challenges and Trends of SSD Design
PDF
Solid State Drives - Seminar for Computer Engineering Semester 6 - VIT,Univer...
PPT
Solid state drives
Solid State Drive (SSD) - SBMathema
Raid
Ssd – Solid State Drive
ssd vs hdd infographic
Raid : Redundant Array of Inexpensive Disks
What is R.A.I.D?
Introduction to Solid State Drives
Solid State Drives - Seminar Report for Semester 6 Computer Engineering - VIT...
SC12 S3700 Series launch event gold
Ssd(solid state drive )
Solid state drive (ssd)
Solid state drive
Trio nas qsan storage
Enterprise SSD: Who is Adopting Them and Why
Understanding Solid State Disk and the Oracle Database Flash Cache (older ver...
Solid state solid state drives
Solid State Drive Technology - MIT Lincoln Labs
Challenges and Trends of SSD Design
Solid State Drives - Seminar for Computer Engineering Semester 6 - VIT,Univer...
Solid state drives
Ad

Similar to Geek Sync | Storage for the DBA (20)

PPT
Storage School 1
PPTX
2015 deploying flash in the data center
PPTX
2015 deploying flash in the data center
PPTX
Spinning Brown Donuts
PPTX
Spinning Brown Donuts: Why Storage Still Counts
PPTX
Information storage and management
PPTX
Deploying ssd in the data center 2014
PPT
Storage Managment
PPTX
04.01 file organization
PPTX
London VMUG Presentation 19th July 2012
PPTX
It's the End of Data Storage As We Know It (And I Feel Fine)
PDF
Josh Krischer - How to get more for less (4 november 2010 Storage Expo)
PDF
Z4R: Intro to Storage and DFSMS for z/OS
PDF
Z109889 z4 r-storage-dfsms-jburg-v1909d
PDF
Operating System File Management disk_management.pdf
PDF
Storage Networking and Overview ppt.pdf
PPTX
Gives an overview of intelligent storage system
PPTX
IO Dubi Lebel
PDF
Insiders Guide- Managing Storage Performance
PPTX
9_Storage_Devices.pptx
Storage School 1
2015 deploying flash in the data center
2015 deploying flash in the data center
Spinning Brown Donuts
Spinning Brown Donuts: Why Storage Still Counts
Information storage and management
Deploying ssd in the data center 2014
Storage Managment
04.01 file organization
London VMUG Presentation 19th July 2012
It's the End of Data Storage As We Know It (And I Feel Fine)
Josh Krischer - How to get more for less (4 november 2010 Storage Expo)
Z4R: Intro to Storage and DFSMS for z/OS
Z109889 z4 r-storage-dfsms-jburg-v1909d
Operating System File Management disk_management.pdf
Storage Networking and Overview ppt.pdf
Gives an overview of intelligent storage system
IO Dubi Lebel
Insiders Guide- Managing Storage Performance
9_Storage_Devices.pptx
Ad

More from IDERA Software (20)

PPTX
The role of the database administrator (DBA) in 2020: Changes, challenges, an...
PPTX
Problems and solutions for migrating databases to the cloud
PPTX
Public cloud uses and limitations
PPTX
Optimize the performance, cost, and value of databases.pptx
PPTX
Monitor cloud database with SQL Diagnostic Manager for SQL Server
PPTX
Database administrators (dbas) face increasing pressure to monitor databases
PPTX
Six tips for cutting sql server licensing costs
PDF
Idera live 2021: The Power of Abstraction by Steve Hoberman
PDF
Idera live 2021: Why Data Lakes are Critical for AI, ML, and IoT By Brian Flug
PDF
Idera live 2021: Will Data Vault add Value to Your Data Warehouse? 3 Signs th...
PDF
Idera live 2021: Managing Digital Transformation on a Budget by Bert Scalzo
PDF
Idera live 2021: Keynote Presentation The Future of Data is The Data Cloud b...
PDF
Idera live 2021: Managing Databases in the Cloud - the First Step, a Succes...
PDF
Idera live 2021: Database Auditing - on-Premises and in the Cloud by Craig M...
PDF
Idera live 2021: Performance Tuning Azure SQL Database by Monica Rathbun
PPTX
Geek Sync | How to Be the DBA When You Don't Have a DBA - Eric Cobb | IDERA
PPTX
How Users of a Performance Monitoring Tool Can Benefit from an Inventory Mana...
PPTX
Benefits of Third Party Tools for MySQL | IDERA
PPTX
Achieve More with Less Resources | IDERA
PPTX
Benefits of SQL Server 2017 and 2019 | IDERA
The role of the database administrator (DBA) in 2020: Changes, challenges, an...
Problems and solutions for migrating databases to the cloud
Public cloud uses and limitations
Optimize the performance, cost, and value of databases.pptx
Monitor cloud database with SQL Diagnostic Manager for SQL Server
Database administrators (dbas) face increasing pressure to monitor databases
Six tips for cutting sql server licensing costs
Idera live 2021: The Power of Abstraction by Steve Hoberman
Idera live 2021: Why Data Lakes are Critical for AI, ML, and IoT By Brian Flug
Idera live 2021: Will Data Vault add Value to Your Data Warehouse? 3 Signs th...
Idera live 2021: Managing Digital Transformation on a Budget by Bert Scalzo
Idera live 2021: Keynote Presentation The Future of Data is The Data Cloud b...
Idera live 2021: Managing Databases in the Cloud - the First Step, a Succes...
Idera live 2021: Database Auditing - on-Premises and in the Cloud by Craig M...
Idera live 2021: Performance Tuning Azure SQL Database by Monica Rathbun
Geek Sync | How to Be the DBA When You Don't Have a DBA - Eric Cobb | IDERA
How Users of a Performance Monitoring Tool Can Benefit from an Inventory Mana...
Benefits of Third Party Tools for MySQL | IDERA
Achieve More with Less Resources | IDERA
Benefits of SQL Server 2017 and 2019 | IDERA

Recently uploaded (20)

PDF
T3DD25 TYPO3 Content Blocks - Deep Dive by André Kraus
PDF
How to Migrate SBCGlobal Email to Yahoo Easily
PDF
PTS Company Brochure 2025 (1).pdf.......
PDF
Which alternative to Crystal Reports is best for small or large businesses.pdf
PDF
Upgrade and Innovation Strategies for SAP ERP Customers
PDF
SAP S4 Hana Brochure 3 (PTS SYSTEMS AND SOLUTIONS)
PDF
Why TechBuilder is the Future of Pickup and Delivery App Development (1).pdf
PDF
Addressing The Cult of Project Management Tools-Why Disconnected Work is Hold...
PPTX
Reimagine Home Health with the Power of Agentic AI​
PDF
Adobe Illustrator 28.6 Crack My Vision of Vector Design
PPTX
Computer Software and OS of computer science of grade 11.pptx
PDF
Internet Downloader Manager (IDM) Crack 6.42 Build 41
PDF
Nekopoi APK 2025 free lastest update
PDF
2025 Textile ERP Trends: SAP, Odoo & Oracle
PDF
Designing Intelligence for the Shop Floor.pdf
PPTX
Transform Your Business with a Software ERP System
PPTX
Oracle E-Business Suite: A Comprehensive Guide for Modern Enterprises
PDF
EN-Survey-Report-SAP-LeanIX-EA-Insights-2025.pdf
PDF
Navsoft: AI-Powered Business Solutions & Custom Software Development
PDF
System and Network Administraation Chapter 3
T3DD25 TYPO3 Content Blocks - Deep Dive by André Kraus
How to Migrate SBCGlobal Email to Yahoo Easily
PTS Company Brochure 2025 (1).pdf.......
Which alternative to Crystal Reports is best for small or large businesses.pdf
Upgrade and Innovation Strategies for SAP ERP Customers
SAP S4 Hana Brochure 3 (PTS SYSTEMS AND SOLUTIONS)
Why TechBuilder is the Future of Pickup and Delivery App Development (1).pdf
Addressing The Cult of Project Management Tools-Why Disconnected Work is Hold...
Reimagine Home Health with the Power of Agentic AI​
Adobe Illustrator 28.6 Crack My Vision of Vector Design
Computer Software and OS of computer science of grade 11.pptx
Internet Downloader Manager (IDM) Crack 6.42 Build 41
Nekopoi APK 2025 free lastest update
2025 Textile ERP Trends: SAP, Odoo & Oracle
Designing Intelligence for the Shop Floor.pdf
Transform Your Business with a Software ERP System
Oracle E-Business Suite: A Comprehensive Guide for Modern Enterprises
EN-Survey-Report-SAP-LeanIX-EA-Insights-2025.pdf
Navsoft: AI-Powered Business Solutions & Custom Software Development
System and Network Administraation Chapter 3

Geek Sync | Storage for the DBA

  • 1. Storage for the DBA Joey D’Antoni February 13, 2018
  • 2. Joey D’Antoni Joey has over 20 years of experience with a wide variety of data platforms, in both Fortune 50 companies as well as smaller organizations He is a frequent speaker on database administration, big data, and career management MVP, MCSE BI and Data Platform VMWare vExpert He is the co-president of the Philadelphia SQL Server User’s Group He wants you to make sure you can restore your data
  • 3. Agenda Storage Terminology Array Cache Setup RAID Types Tiered Storage Disk Alignment Spindle Types Physical Array Diagram
  • 4. Storage Terminology LUN = Logical Unit Number Host = The Server or Servers a LUN is presented to SAN = Storage Area Network Fabric = Fibre network which makes up the SAN Array = Box with the Spindles in it
  • 7. Storage Terminology Disk = How the OS sees a LUN when presented Spindle = Physical disks in the Storage Array IOps = Physical Operation To Disk Sequential IO = Reads or writes which are sequential on the spindle Random IO = Reads or writes which are located at random positions on the spindle
  • 8. Array Cache Setup OLTP databases make poor use of SAN read cache OLAP databases make good use of SAN read cache Try reducing read cache and increasing write cache OLTP databases with high buffer cache hit ratios may be able to have the read cache disabled There is no one correct setup. Every system is different.
  • 9. When Write Cache Gets Full? Doesn’t flush to disk until low watermark is hit Force flushes once high watermark is hit Force flushing completely empties write cache Force flushing disables write cache until write cache is disabled If adjustable, set low very low, and high watermark very high (20/90)
  • 10. Oh No, the power went out! Most arrays have internal batteries Write cache is flushed to disk After flush array powers down On power-up flushed cached is read and committed to LUNs before LUN is made available Make sure SQL is down before the array flushes http://www.flickr.com/photos/robertfrancis/352039299/
  • 12. Spinning Disk and Hybrid Arrays
  • 13. RAID 0 Straight Stripe No redundancy Very fast Low cost Any disk failure looses data Can not survive no disks failures Requires 2 or more disks
  • 14. RAID 1 Full Mirror of data No performance Benefit High Cost Requires 2 disks Can survive 1 disk failure
  • 15. RAID 0+1 Drives Striped, then Mirrored High Cost High Performance Requires 2 disks Can survive 1 spindle failure Requires even number of disks May survive multiple spindle failure
  • 16. RAID 10 (1+0) Drives Mirrored, then Striped High Cost High Performance Requires 2 disks Can survive 1 spindle failure Requires even number of disks May survive multiple spindle failure
  • 17. RAID 5 Low Cost Requires 3+ disks Stripe with Single Parity Can survive a single drive failure Write Penalty Good Performance
  • 18. RAID 6 Stripe with 2 parity Higher cost per gig than RAID 5 Requires 4+ disks Can survive 2 disk failures Similar penalty as RAID 5
  • 19. Not all RAID 5 is Created Equal RAID 6 can be done this way as well.
  • 20. Disk Alignment An old problem, that can still be an issue By default old Windows disks are offset incorrectly giving a 50% performance penalty Correcting this can improve SQL disk performance up to 100% • (64 1k blocks/64k IO)=100% of IO is impacted Must be done before data is put on the disk Windows 2000 - Diskpar Windows 2003 - Diskpart Windows 2008+ - Automatic
  • 22. Auto-Tiering Storage Used in hybrid arrays Gives best performance storage for some types of data Puts hot data SSD Puts idle data on spinning disks
  • 23. SSDs Are Fast, But Different Random I/O becomes a good thing Compressing data can be done by the array, very efficiently All flash SSDs tend to have a different platform architecture These arrays are disruptive to the market in ways beyond pure I/O performance
  • 26. Storage in the Cloud
  • 27. Ten Tips to Healthy Storage Be honest with your storage team Give back storage when you don’t need it, or know you won’t need it for a while If there’s a performance problem, make sure it isn’t your database first It helps if your storage admin understands databases SSDs require different configuration—understand your array
  • 28. Ten Tips to Healthy Storage Always compress your backups (you need to fight this one) Advanced technologies like NMDimm are good for high-end workloads More memory is always better You can’t overcome terrible storage performance The cloud is different, and you may be more in charge

Editor's Notes

  • #7: With block storage, files are split into evenly sized blocks of data, each with its own address but with no additional information (metadata) to provide more context for what that block of data is. You’re likely to encounter block storage in the majority of enterprise workloads; it has a wide variety of uses (as seen by the rise in popularity of SAN arrays). Object storage, by contrast, doesn’t split files up into raw blocks of data. Instead, entire clumps of data are stored in, yes, an object that contains the data, metadata, and the unique identifier. There is no limit on the type or amount of metadata, which makes object storage powerful and customizable. Metadata can include anything from the security classification of the file within the object to the importance of the application associated with the information. Anyone who’s stored a picture on Facebook or a song on Spotify has used object storage even if they don’t know it. In the enterprise data center, object storage is used for these same types of storage needs, where the data needs to be highly available and highly durable. However, object storage generally doesn’t provide you with the ability to incrementally edit one part of a file (as block storage does). Objects have to be manipulated as a whole unit, requiring the entire object to be accessed, updated, then re-written in their entirety. That can have performance implications. Another key difference is that block storage can be directly accessed by the operating system as a mounted drive volume, while object storage cannot do so without significant degradation to performance. The tradeoff here is that, unlike object storage, the storage management overhead of block storage (such as remapping volumes) is relatively nonexistent.
  • #25: Compare the benefits and disadvantages of NVDIMM/Memory1 to NVMe.  NVMe or Non-Volatile Memory Express, sits on the PCIe bus (it’s not strictly a bus but a point-to-point connection).  PCIe is connected on the northbridge, so is closer to the processor.  Because it’s not a bus, PCIe doesn’t have the issues of a shared architecture and so can scale better.  Performance improves with each generation of PCIe and by aggregating channels or lanes together, a single device can deliver more throughput.  So flash devices that use NVMe can deliver high performance and low latency persistent storage.  Intel use NVMe for Optane drives, which have latencies as low as 10µs.  Standard flash-based NVMe drives can deliver performance in line with technologies like Memory1.  Don’t forget, many NVMe drives (in 2.5″ drive form factor) are also hot swappable.  It’s certainly not possible to hot-swap DIMMs.  NVMe flash drives also offer high capacity, compared to what was offered with Memory1. NVDIMM or non-volatile DIMM is a technology that uses the DRAM DIMM form factor and is directly plug compatible with system memory.  As the name suggests, the contents of the DIMM are not lost when the power is turned off.  Diablo calls this technology Memory Channel Storage, with the rest of the industry generally knowing as Storage Class Memory or SCM.  If we compare traditional (for example flash) storage to DRAM, one big difference is the way in which storage is addressed.  DRAM is byte-addressable, whereas flash is block-addressable.  Flash requires an entire block of data to be modified for an update – DRAM can do this at the byte level.