Inteligentné diskové
systémy NetApp
a VMware
STORYFLEX
2010
NetApp advantage of Unified Storage
FAS Unified Storage Family: SAN, NAS, iSCSI
®
Data ONTAP
Compatible scalability
Simplified administration
Superior utilization
Lower cost & complexity
2
NetApp Software Options Features:
Included with base system
Optional feature
43 Advanced Software Features * not available with gateways
1. Data ONTAP 16. CIFS Protocol 35. SnapManager for SQL
2. iSCSI 17. NFS Protocol 36. SnapManager for Exchange
3. iSCSI Host Attach Kit for 18. HTTP Protocol 37. SnapManager for Oracle
AIX, Windows & Linux 19. FCP Protocol 38. SnapManager for SAP
4. FTP 20. Cluster Failover 39. SnapValidator
5. Snapshot 21. FlexClone 40. Operations Manager Core
6. SnapMover 22. MultiStore 41. Operations Manager SRM
7. SyncMirror 23. SnapMirror 42. Protection Manager
8. FilerView 24. SnapRestore 43. Advanced Single Instance Storage
9. FlexVol 25. SnapVault
10. FlexShare 26. Open Systems SnapVault
11. SecureAdmin 27. SnapLock Enterprise
12. AutoSupport 28. LockVault Enterprise
13. Disk Sanitization 29. SnapLock Compliance*
14. RAID-DP* 30. LockVault Compliance*
15. RAID-4* 31. SnapDrive for Windows, Unix & Linux
32. NearStore feature
33. MetroCluster*
34. Single Mailbox Recovery
3 3
Flexible deployment of
multiprotocol, unified storage
BLOCK SERVICE FILE SERVICE
Enterprise Departmental Enterprise Departmental
iSCSI
Fibre NAS over Corporate
Channel dedicated LAN
Ethernet
SAN NAS
(Block) (File)
IP FCP FAS, NearStore, V-Series
4
FAS: One scalable architecture
Unified Storage Architecture
FAS6080
FC-SAN, iSCSI, NAS
Scalable Performance and Capacity
Data ONTAP® provides single FAS6040
application interface FAS3170
FAS3160 1,176TB
FAS3140
Performance
840TB
FAS2050
840TB
FAS2020 672TB
420TB
104TB
68TB Train once for all of your storage!
Simple, pain-free upgrades!
Data ONTAP®
5
V-Series Products Overview
Transform Your Current Storage
NetApp® V-Series virtualizes SAN storage from multiple vendors into a single pool
for file or block access to provide the benefits of Data ONTAP® and the
NetApp manageability software family.
V6080
V6040
V3170
V3160
V3140 1,176TB
840TB
840TB
672TB
420TB
Supporting Disk Arrays from these Storage Vendors
SSD
6
Single Architecture means Seamless Scaling
• Seamless Scalability
1.176
TB
• Zero Data Migration
FAS6000
• Investment protection V Series
FAS3100
V Series
*
FAS2050
Data ONTAP
FAS2020
68 TB
* Disk shelf conversion
7
FAS Primary Storage Product
Comparison
FAS6080 FAS6040 FAS3170 FAS3160 FAS3140 FAS2050 FAS2020
Max. capacity 1176TB 840TB 840TB 672TB 420TB 90TB 68TB
Max. HDD 1176 840 840 672 420 104 68
Main memory 64GB 32GB 32GB 16 GB 8 GB 4 GB 2 GB
NV RAM 4GB 1GB 4GB 4 GB 1 GB 512 MB 256 MB
Gbit Ethernet 12 x 1Gb 12 x 1Gb 4 x 1 Gb 4 x 1Gb 4 x 1Gb 4 x 1 Gb 4 x 1Gb
Onboard ports
Gbit max ports 52 52 36 36 36 8 4
Fibre Channel 16 16 8 8 8 4 4
Onboard ports
FC max ports 56 56 40 40 40 8 4
PCI expansion 10 PCIe 10 PCIe 8 PCIe 8 PCIe 8 PCIe 2 PCIe ––
slots 6 PCI-X 6 PCI-X
8
SnapShot
SnapRestore
Snapshot™:
Why NetApp is the Technology Leader
up to 255 Snapshots per volume
no performance impact
NOT the common „copy-on-write“
best space efficiency
only changed BLOCKS are saved
no license cost
Integral part of DataOntap OS
full integration into OS and Database
with SnapManager and SnapDrive software
Snapshot is a trademark of Network Appliance Inc. 10
Copy-out Other vendors
EMC SnapView (CLARiiON), EMC TimeFinder
(Symmetrix/DMX), SnapSure (Celerra and NS600)
HPQ Vsnap (EVA 3000, 5000) and Business Copy (EVA 5000,
XP, VA), Virtual Replicator (MSA), Persistent Storage
Manager (e7000, b2000, b3000, b1000)
Microsoft VolSnap (ships with Windows Storage Server’s VSS
facility) and Persistent Storage Manager (SAK 2000)
VERITAS FlashSnap (space saving Storage Checkpoints), VxVm,
VxFS
IBM FlashCopy (TotalStorage DS4000 series arrays)
Dell SnapView (EMC CLARiiON CXx00 products),
ActiveArchive (PowerVault NAS products)
Sun Point-in-Time software, formerly Instant Image, (StorEdge
products), UFS snapshots (bundled in Solaris 9)
11
NetApp Snapshot™ Technology
Blocks in Blocks
LUN or File on the Disk
Take snapshot 1
– Copy 4k-pointers
A
A A only
B
B B – No data movement
C
C C
Snap
Pointer 1
12
NetApp Snapshot™ Technology
Blocks in Blocks
LUN or File on the Disk
Take snapshot 1
A
A A Continue writing data
B1
B
B1 B
C
C C
Take snapshot 2
B1 – Copy 4k-pointers only
– No data movement
A A
B B
C C
Snap Snap
2
Pointer 1 Pointer 1
13
Restore Snapshot™
Blocks in Blocks
LUN or File on the Disk
snapshot 1 and 2
exists
A
A A
B1
B
B1
B B restore snapshot 1
C
C C
– Copy back pointers
B1
– No data movement
takes only seconds
A A – Independent of size
B B1
C C
Snap Snap
Pointer 1 Pointer 2
14
Others: “copy-on-write” method
Blocks in Blocks
LUN on the Disk
Take snapshot 1:
– Create copy out region 1
A A – Create pointers to old
B B blocks and copy out
C C
Copy
Out 1
A
B
C
Snap 1
15
Others “copy-on-write” method
Blocks in Blocks
LUN on the Disk
Take snapshot 1
Continue writing data
A A
– Block changes
B
B1 B
– Read old block; write to
C C
copy out region
Copy – Update snap pointer to
Out 1
copy out region
– Update block on disk
A
B One write requires:
C – 1 read (old data)
– 1 write (old data)
– 1 write (new data)
Snap 1
3x overhead
16
NetApp Snapshot™ Advantages
Others with
NetApp copy-on-write Highlights of NetApp Technology:
A A
B B1 No data movement
Used
C Disk C2 – guarantees performance
Space
B1 B – enables more snapshots
C2 C
– allows fast restore
C
Absolute minimum overhead
– guarantees disk space efficiency
Side-by-side comparison
after two snapshots
17
SnapShot Integration
for OS and Databases
SnapManager Oracle SnapDrive
Host OS
Database
SnapManager SQL OpenSSV
SnapManager Exchange
SnapVault
FAS / Nearstore / V-Series
FlexClone
Data OnTap
SnapMirror
SnapRestore
SnapShot
WAFL / NVRAM / RAID4 / RAID-DP / FlexVol
19
FlexVol™
FlexClone™
FlexVol™ Maximizes Efficiency
of Storage Consolidation
Traditional Industry Model With NetApp FlexVol
• Volumes “co-located” • Single shared pool of storage
• Volumes tied to physical disk • Higher utilization
• Issues: Utilization, management • Dynamically reconfigure
Vol 1 Vol 2
Vol 3
Vol 2 Vol 1
Vol 4 Vol 4
Free
Vol 3
21
Improved Performance
Regular volume
• Volume performance limited
• “Hot” volumes isolated
FlexVol
• Shares spindle performance
• Typically >2x improvement
(10-disk vol. vs. shared 40-disk aggregate)
22
Application Development and Testing
Example without FlexClones
Production Mirrored Copy
Production database 100GB
Mirror copy 100GB
Development copies 300GB
Testing copies 300GB
Total: 800GB
Dev 1 Dev 2 Dev 3
8x actual storage requirement
Time consuming
Resource overhead
Test 1 Test 2 Test 3
23
FlexClone™
Improves Flexibility and Efficiency
User reads FlexClone User
unchanged user writes Instant, Snapshot-based logical r-w copy
Snapshot
data in base Only new data consumes additional storage
in the FlexClone or the base flexible volume,
as appropriate
FlexClone
(read-write) Usage examples
– Testing using large production databases
– Product development, with numerous
variations on a core design
– Concurrent simulations (e.g., crash tests)
Snapshot – Data mining where R/W required
(read-only)
– Parallel processing
New
– Large system deployments where small
production
writes variations required, e.g., grid computing
Key Benefit
Base flexible volume – Dramatic cost savings vs. competitive
alternatives
24
Application Development and Testing
Example
Production Mirrored Copy
Production database 100GB
Mirror copy 100GB
Development copies 30GB
Testing copies 30GB
Total: 260GB
Dev 1 Dev 2 Dev 3
Over 67% reduction in storage
required
Near instantaneous copies
Test 1 Test 2 Test 3 Negligible overhead
Ability to have lot more test and
dev copies
Assumption: up to 10% change in data in the test and dev environments
25
RAID-DP
Dual Parity RAID
RAID-DP:
cost of RAID 6 - performance of RAID 10
What is it:
RAID Level protecting against ALL double disk failures
physical disk utilization equal to Parity RAID
much higher performance than classic RAID-6
write performance even higher then competition´s RAID-10
Why it is important:
NOT because two disks fail at the same time
nowadays disks have very high bad block rate
to avoid data loss If a bad block is discovered during RAID
reconstruct
RAID-DP reduces the risk of Data Loss compared to
RAID-5 dramatically!
28
RAID-DP: Performance Comparision
Veritest Performance Test on a 400GB LUN
OLTP Workload: 60% Random Read, 40% Random Write
IOPS 14100 IOPS
7230 IOPS
CX3-80 FAS3070
32 Disks RAID-10 28 Disks RAID-DP
Striped MetaLUN LUN
Because of DataOntap´s intelligent WAFL Technology!
29
WAFL®
WAFL: The performance boost
Look at the numbers:
CX3-80 LUN delivers 7230 random IOPS
– this is approximately what 32 Disk are able to deliver
with random IO (32 x 200)
FAS3070 LUN delivers 14100 random IOPS
– this is more than the disks (28) would ever be able to
deliver in random access mode
WAFL: Write Anywhere File Layout
internal Datastructure of NetApp Systems
Implements special Write Cache destage algorithm
WAFL is base for NetApp SnapShot Technology
31
Small Random Write Acceleration
WAFL translates „small random writes“ into
„large sequential writes“ !
Host IO Queue Write Cache Disk Performance Principle:
Example:
1.) Write one 256k block to a
disk takes nearly the same
time as writing on 8k block
2.) Write one 256k block is
approx. 30 times faster then
32 x 8k
WAFL Filesystem
on Disks
32
WAFL: Write Anywhere File Layout
NetApp’s patented filesystem for Improved Performance
Berkeley Fast File System/Veritas File System/NTFS/etc. –
Writes to pre-allocated locations (data and metadata)
...
WAFL – Writes to nearest available free block, thus reducing
disk seeking (the #1 performance challenge when using disks)
1-2 MB
Cylinders
...
33
LUN Thin
Provisioning
Thin Provisioning
Allows to „oversubscribe“ physical diskspace
– LUNs and Volumes do not contain data immediatly
when created
– Typically filled with data over time
– Physical diskspace is really needed when data is
written to the LUN
NetApp Technology allows customers to buy storage
when it is really needed
35
Classic space allocation
Typical LUN Provisioning:
1TB Total usable
Space
– 1TB RAID-group
– Space preallocated with every Free Space
created LUN
– 3x 200GB LUN 400GB free
200GB LUN 3
– much less space used by real
data: e.g. only 300GB 200GB LUN 2
– 300GB Free space is “blocked”
200GB LUN 1
36
DataOntap can distinquish
“Data” and “LUN allocation Size”
Thin Provisioning Example:
1TB Total usable
Space
– 1TB RAID-group
– LUNs can be created without Free Space
preallocating space
– Space is consumed when
200GB LUN 3
LUNs are filled with data
e.g. 300GB 200GB LUN 2
– 700 GB available for shared
use 200GB LUN 1
37
Oversubscription
– LUN creation is not limited to
physical available space
– This requires special
Free Space
monitoring features
– Space consumed by Data
reaches defined Threshold 200GB LUN 3
– Expand physical Space
200GB LUN 2
(auto_grow Volumes and expand Disk
Aggregates online as required)
200GB LUN 1
38
A-SIS
Deduplication
A-SIS Defined
A-SIS: Advanced Single Instance Storage
– Official marketing name is “A-SIS deduplication”
Deduplicates 4K blocks anywhere in the active file system of a
flexvol on a NearStore system
Runs as background process
Application Transparent
– Context independent block sharing; requires no prior
knowledge of block relationships
Initial positioning is for data retention / archive
Runs on secondary storage AND primary!
40
A-SIS Deduplication: Function
Data
Meta Data
Deduplication
Process
Advanced
Single Instance Storage
Meta Data
General data in
flexible volume
A-SIS data in
flexible volume
41
Typical Space Savings Results
in archival and primary storage environments:
Video Surveillance 1%
PACS 5%
Movies 7%
Email Archive 8%
ISOs and PSTs 16%
Oil & Gas 30%
Web & MS Office 30 - 45%
Home Dirs 30 - 50%
Software Archives 48%
Tech Pubs archive 52%
Virtualized environments 50 - 80%
SQL 70%
In data backup environments, space savings can be much higher. For instance,
tests with Commvault Galaxy provided a 20:1 space reduction over time,
assuming daily full backups with 2% daily file modification rate. (Reference:
http://www.netapp.com/news/press/news_rel_20070515)
42
PAM module –
Do you really
need SSD drives?
Performance Acceleration Module
Adds new performance improvement capabilities
Keep up with performance needs without
increasing drive counts
Extend capability of an existing system
Separate performance from capacity
Use less power
Use less space
Use less cooling
44
Performance Acceleration Module
45
MetroCluster
Metropolitan area network
active-active cluster
with RAID1
Clustered Failover [Normal Operation]
Aggregate 2
Fibre Channel
Controller A Aggregate 1 Controller B
NVRAM =
Cluster Interconnect Battery-
NVRAM NVRAM protected
- Heartbeat journal of
- NVRAM mirroring operations
IP or FC
fabric
Windows & UNIX Clients Windows & UNIX Clients
47
Clustered Failover [Takeover]
Aggregate 2
Controller A Aggregate 1 Controller B
Controller B
Cluster Interconnect
NVRAM NVRAM
IP or FC
fabric
Windows & UNIX Clients Windows & UNIX Clients
48
SyncMirror - Synchronous Mirroring
Data Center
Local SyncMirror
– Mirror on the „Backend“
– Two synchronous mirrors
within a single volume
vol X
– Both plexes are updated vol Y
synchronously on writes. vol X’
49
Stretch MetroCluster – Campus
distances
Building A Similar to local cluster Building B
configuration, but with
long cables
FAS2 Heartbeat, NVRAM sync’ing
V-Series3
vol X vol Y
vol Y’
Benefits vol X’
Disaster protection
Complete redundancy
Cluster interconnect Up-to-date mirror
A-loop Site failover
B-loop
Reciprocal
Independent reads
50
Fabric MetroCluster – Long Distance
(>500 m to 100 km)
Building Building
A B
NAS/SAN
Heartbeat,
NVRAM sync’ing,
FC over dark fibre
DWDM DWDM
Benefits
Disaster protection
vol X vol Y’ Complete redundancy vol X’ vol Y
Up-to-date mirror
Site failover
Cluster interconnect Reciprocal
Independent reads
MetroCluster and SynchMirror combines to the fastest
full synchronous Disaster Recovery Concept 51
SnapMirror ®
Remote Replication
for Disaster recovery
SnapMirror®: Disaster recovery for
homogeneous NetApp deployments
• Mirrors volumes or single directories to remote appliance
Disaster recovery • Includes 3 mirroring options, depending on data criticality,
Centralized backup budget, distance: Synch, asynch, or semi-synch
Data migration/distribution • Source and target needn’t be identical (asymmetric DR)
• IP and FC fabric-attached clients
IP and/or FC active-active mirroring path
FAS / V-Series / NearStore
• 1-to-1, many-to-1, 1-to-many, cascading, reciprocal
sources
• Targets can be mined, backed up
• Auto-failover of clients with VFM® (Virtual File Manager)
Data center
• Optimized failback based on last good Snapshot
FAS / V-Series / NearStore
Vol1
target(s)
DR center
LAN/WAN Vol1
SnapMirror®
Directory ABC Directory
ABC
Remote office
53
SnapVault®
Block Incremental backup
SnapVault® Topology
Data Center DAS / networked filesystems on Remote Office
Files / LUNs on Linux UNIX Windows
FAS / V-Series OSSV
agent
SV OSSV OSSV OSSV SV
primary agent agent agent Hourly or daily, primary
block-
Hourly or daily,
Hourly or based xfers
file-based xfers
daily, block- Network
based xfers Network WAN
Backup
Server DR Center
Open Systems Weekly/Monthly Backup
Networked Full Backup Server
Storage Baseline, control
Data
then
changed
blocks Tape Library
SnapVault
“secondary” stored Data
WAN
Weekly/Monthly
SnapMirror NearStore Full Backup
NearStore
= File-level transfers
SnapVault supports filesystems native to various Windows, UNIX, and Linux operating systems
= Block-level transfers after baseline
55
NetApp - Best
Storage for Virtual
Infrastructures
Where Can NetApp Help You?
Application Integration:
Business Apps Messaging Technical Apps
& Database & Collaboration & File Services
Infrastructure:
Consolidation & Virtualization
Data Protection and Retention
Flexibility and New Deploy with
efficiency capabilities confidence
Go further, faster ™
57
57
No-Compromise Availability
Exceeding 5 9’s availability
Autosupport data from IDC-validated
15,000+ customer systems
100,0000
Six 9s 99,9999 FAS6xxx HA
99,9998 FAS3xxx HA
99,9997
FAS2xxx HA
99,9996
99,9995 FAS2xx HA
99,9994 FAS9xx HA
99,9993
99,9992
99,9991
Five 9s 99,9990
4.07 5.07 6.07 7.07 8.07 9.07 10.07 11.07 12.07 1.08
FAS 6-Month Rolling Average Availability
IDC White Paper: FAS Storage Systems: Laying the Foundation for Application Availability
58
58
Accelerate Your Application Dev & Test
Traditional Provisioning NetApp Provisioning
27 minutes 3 seconds
Time-consuming Instant copies with
physical copies FlexClone®
Slow to provision or Quick provisioning
reprovision with FlexVol®
Manual and complex
Simple commands or
policies
59
59
Reduce Data Management
App Admin DBA Server Admin Faster, better
response to business
needs
More productive
Automated Services people
•Provision•Backup/Recover Enabling technologies
•DR •Test/Dev Copies
Provisioning
Manager
Policies
Protection Manager
Policies
•Create new
policies
•Copy policies
•Repurpose
60
policies 60
Store More, Buy Less
Outstanding space efficiency
Eliminate duplicate data
– 20:1 reductions with
deduplication
Stop duplicating unchanged data
– NetApp's copies use a fraction
of the space
Make your storage work harder
– Thin provisioning yields
higher utilization
Protect your data without
copy bloat 61
61
Efficiency Is in the Software
Benefits compound when using multiple features
Save Snapshot™ Copies Save Deduplication
over Point-in-time copies that write up to Removes data
80% only changed blocks. No 95% redundancies in primary
performance penalty.
and secondary storage.
Save Virtual Copies Save RAID 6 Protection
up to
over (FlexClone®) 46% (RAID-DP®)
80% Protects against double disk
Near-zero space, instant
“virtual” copies. Only failure with no performance
subsequent changes in penalty.
cloned data set get stored.
Save Thin Provisioning
up to
33% (FlexVol®)
Create flexible volumes that
appear to be a certain size but
are really a much smaller pool.
62
62
Deduplication
Pervasive Deduplication
20:1 reduction in
secondary
30-50% reduction in
primary
Capability, not a platform
Before After
66
66
Calculate the Efficiency Savings
NetApp® Traditional Arrays
22 Disks
9
14 Disks
16
11 61 Disks
28
14 Disks
33
Primary Storage
Double Disk
Data Protection
Deduplicate
Primary Data
Add VMware® Disk-
Based Backup
Add VMware SRM
Storage Virtualization should drive cost reductions
68
68
MetroCluster Provides Continuos Availability
and Fault-Tolerance for VMware Environment
69
Simplifying File Services
Solution Highlights
Time-to-Market Improved Storage
Acceleration Productivity Efficiency
Multiprotocol support PAM, SAA
Leading performance Snapshot™ Deduplication
Improve productivity of your end users and IT staff with fast global access to data
– Accelerate time to market using intelligent caching with Performance Acceleration
Module and Storage Acceleration Appliance and multiprotocol (CIFS and NFS) support
Double storage utilization while using 50% less storage
®
– Eliminate redundant data with FlexVol , space and performance efficient copies and
deduplication
Improve data protection while reducing disruptions by 50% or more
– Create frequent, quick backups or easily replicate data and enable easy user-driven
® ® ®
restores with Snapshot, SnapVault , SnapRestore , and SnapMirror
Partnering with key engineering application and management vendors
– Joint testing and escalation, industry leading certifications for engineering applications
giving a solution that works right out of the box
70
NetApp Confidential - Internal Use Only 70
Server Virtualization & NetApp:
The Perfect Match
Server Virtualization NetApp ®
Improves:
Accelerates
– Cost of ownership
deployment
– Manageability
– Responsiveness Uses 90% less storage
– Utilization Superior manageability
Creates Solves problems
challenges: with:
– Backup – Backup
– Disaster recovery – Disaster recovery
– Copy proliferation – Copy proliferation
71
71
The Best Storage for Virtual
Infrastructures
Solution Highlights
Storage Storage Instant Backup/
Efficiency Flexibility Restore
VM1
VM1
VM1
Deduplication SnapManager® Snapshot™
Thin provisioning for Virtual Infrastructure SnapRestore®
Reduce storage for virtual infrastructures by 50% or more
– Eliminate redundant VM data with deduplication and improve
storage utilization with thin provisioning
Back up and recover VMs in seconds
– Free up your virtual server host with backups built directly into
the storage
Provision storage as fast as virtual servers
– Enable your virtual server administrators to provision storage
and clone VM instantly
72
72
Rastislav Maniak
rmaniak@storyflex.cz