Transcript Slide 1

Oracle and NetApp
Mike McGuiggan
Consulting Systems Engineer
Northeast Region
Oracle and Network Appliance®
12 years of innovation driving down infrastructure costs
AND increasing enterprise IT productivity
2005
NetApp deployment at
Oracle:
- over 3200 TB
- 14,000 Servers
ONTAP 7G FlexVol &
FlexClone deployed
2001
NetApp runs its core
business on Oracle
E-Business Suite
1999
NetApp helps
establish
1995
Oracle Storage
First filer Compatibility
for Oracle Program (OSCP)
DB sold
NetApp and Oracle
Blueprints from
Oracle Consulting
2003
NetApp runs Oracle ERP,
3000+ Oracle on
CRM and HR
NetApp customers
SnapValidator announced
in production
Linux competency EM Grid Control Integration
1997
Oracle
certifies
NetApp
Oracle B2B
Exchange selects
NetApp
NetApp is Certified
Solution Partner
NetApp deployment at
Oracle:
- over 6900 TB
- 20,000+ servers
Oracle Enterprise
Linux 4 Support
SnapLock Integration
center
2006
2004
2002
2000
2007
NetApp deployment at
Oracle: 500TB
Oracle NetApp Joint
Escalation Team
ONJET
Joint Center of
Excellence (CoE) and
professional services
© 2008 NetApp. All rights reserved.
NetApp deployment at
Oracle:
- over 2000 TB
- 9,500 Servers at Oracle
Austin Data Center
Enterprise Grid Alliance
Oracle builds next Apps
on NetApp storage
NetApp deployment at
Oracle:
- over 5500 TB
- 20,000 servers
OCS SnapLock
Integration
SnapManager for
Oracle Launched
5000+ joint customers
2
Safra Catz, President and CFO of Oracle
OOW quote:
“I would like to introduce my good friend Tom
Mendoza…………NetApp is a very, very close
partner of Oracle…..we have worked together for a
very long time in advancing technology, grid
computing, and data center consolidation……..for
any of you that has visited our Austin Data Center, it
is a commercial for NetApp. We voted for them with
our money.”
© 2008 NetApp. All rights reserved.
3
Oracle Austin Data Center.
Over 20,000 servers.
Over 6900+ TB of NetApp Storage.
Austin Data Center Stats
• More than 20,000+ servers
• Largest Dell/Linux installation on earth
• 6,900+ terabytes of NetApp storage
• Largest NetApp single installation on earth
• Scaling at over 60TBs of storage each month
• Over 500 mission-critical customers' apps
hosted on the premises
• Primary data center for 50,000+ Oracle
employees
• 2 acres of data center raised floor space
Results
•
Experiencing 30% higher utilization and faster
application deployment
• Storage management overhead reduced 50%
• Global IT has 88% faster response time on
NetApp FC SAN
© 2008 NetApp. All rights reserved.
4
The Bottom Line
Total Cost Comparison
Typical Database Applications
▶ Modular arrays – 55% to 71% more than NetApp
▶ Frame arrays – over 250% more than NetApp
300
260
Business Cost of
Downtime
Total Cost Index
(Normalized to NetApp San = 100)
250
200
171
+243%
155
150
100
100
+108%
+92%
Internal Operational
Costs
Acquisition &
Ongoing Vendor
Costs
50
0
NetApp FAS
HP EVA
EMC CLARiiON
EMC Symmetrix
Source: Mercer Database Storage Total Cost Comparison Interviews, December 2005 - January 2006.
© 2008 NetApp. All rights reserved.
5
Oracle as a Customer - Highlights
 99% of Oracle application development is on
NetApp (2400TB)
 99% of Oracle OnDemand “Production
Applications and DB” (2100 TB)
 100% Oracle University (120TB)
 100% Oracle Demos DB & Applications
(500TB)
 50% Database Development (300TB)
 65% of Oracle internal production (2420 TB)
“NetApp has reduced the overhead necessary to perform storage-related
management activities by 50%.”
© 2008 NetApp. All rights reserved.
6
NetApp Supports Them All!
Fabric Attached Storage Topology
iSCSI
SAN
Enterprise
SAN
Enterprise
NAS
Dedicated
Ethernet
Fibre
Channel
Departmental
NAS
Corporat
e
LAN
NAS
(File)
SAN
(Block)
NetApp®
FAS
© 2008 NetApp. All rights reserved.
7
Simplifying Database Management:
NetApp’s Value Proposition to the DBA
© 2008 NetApp. All rights reserved.
8
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
9
Database Performance Tuning and
Capacity Planning Challenges
 Storage can be a performance
bottleneck
 Duplicate database copies are
needed to maintain availability
and SLAs
 Performance of each volume is
limited to the number of spindles
within
 “Hot” volumes can’t be “helped”
by disks in other volumes
 Capacity planning, forecasting
and management are imprecise
tasks
 Availability, scalability and
capacity must be maintained or
increased while driving down
costs…
© 2008 NetApp. All rights reserved.
10
Provision
Software – on – Demand … with LESS
FlexVol™ - Flexible, Virtualized Volumes & LUNs
Conventional Volumes & LUNs
vol0
vol1
vol2
vol3
data
P
HM DIR’s
P
SQL “HOT”
P
P
Test
Oracle/Caché
Data ONTAP 7G™ – FlexVols™
Manual, inefficient
Unused space not
available to others
FlexVol™s share disk Aggregate
add disks on-the-fly
WAFL® Provisions-on-Demand P
Grow & shrink non-disruptively
P
n Raid Groups = AGGREGATE = Physical Storage Pool
FlexVol™ = Logical Data Container (1MB – 16TB)
More shared spindles mean NO “HOT” vols / LUNs
© 2008 NetApp. All rights reserved.
free
Full provisioning tied
to physical disks
Higher Storage Utilization
Improved Performance
Thin Provision
11
WAFL®, Snapshot™, and FlexClone™:
The Foundation
Inodes,
block maps
Active
Filesys
or LUN
Snapshot
A
B
C
FlexClone
D
D’
Y
Disk blocks
© 2008 NetApp. All rights reserved.
NetApp’s Unique DNA
 WAFL – Write Anywhere File Layout
– Core block and file storage
services
– Resiliency features (RAID-DP™)
 Snapshot
– Near-instantaneous, point-in-time
“copy” of file system (vol) or LUN
– Read-only
 SnapRestore®
– Near-instantaneous “rollback” of
vol or LUN to prior Snapshot copy
 FlexClone
– Near-instantaneous, WRITABLE
“copy” of vol or LUN
– Same space-sharing
12
Introducing SnapshotTM
After Snapshot
Block
Update
Before
Snapshot
Disk Usage
Root
Inode
A
B
Root
Inode
Snapshot
(Inode Copy)
C
D
C’
Disk Blocks
© 2008 NetApp. All rights reserved.
13
Introducing SnapRestore®
After SnapRestore
Snapshot
Disk Usage
(New) Root
Inode
A
B
Root
Inode
Snapshot™
(Inode Copy)
C
D
C’
Disk Blocks
© 2008 NetApp. All rights reserved.
14
Database Performance Tuning
Example
Test OLTP
Database 2
Test OLTP
Database 1
8-Disk
Aggregate
24-Disk
Aggregate
Transaction Log
Archive Log
Flashback
© 2008 NetApp. All rights reserved.
Data Files
32-Disk
Aggregate
All
Files
15
Database Performance Tuning
Example
Test OLTP
Database 1
20%
Test OLTP
Database - 2
17%
15%
10%
5%
0%
-5%
-10%
8-Disk
Aggregate
24-Disk
Aggregate
-15%
-20%
Transaction Log
Archive Log
Flashback
© 2008 NetApp. All rights reserved.
Data Files
-25%
32 Disk
Aggregate
-24%
All files
Disk Utilization
Transactions/min
16
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
17
Database Backup and Recovery
Challenges
 RMAN backups consume
significant physical resources
 Cold backups lead to lower
SLAs
 Separate backups on each
platform
 Time-to-recover from tape
becomes prohibitive
 DBA’s time spent on non-valueadd backup/restore tasks
 Backups performed with less
frequency
© 2008 NetApp. All rights reserved.
18
Backup and Recovery with Snapshot and
SnapRestore
 Significant time savings
 Stay online
 Reduce system and
storage overhead
 Consolidated backups
 Backup more often
300GB
Database
Primary Backup
Secondary Backup
SnapVault
Time to
Backup
Snapshot
Active
Filesys or
LUN
FlexClone
To Tape (60GB/Hr Best Case)
Time to
Recover
C D D’ Y
A B
Snapshot™
From Tape
Redo Logs
SnapRestore®
Redo Logs
0
1
2
© 2008 NetApp. All rights reserved.
3
4
Time in Hours
5
6
7
8
19
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
20
RAID-DP Reliability
RAID
 2,000 to 4,000 times more
secure than regular RAID
Protects against single
disk failure
P
 More reliable than mirroring for
double-disk failure
RAID DP
Protects against any
P
two-disk failure
DP
 14% Parity Overhead vs.
50% Overhead w/Mirror(*)
 72% more usable capacity
than competitive offerings(*)
(*) Comparing 2P+12D vs. 7+7 Mirror
© 2008 NetApp. All rights reserved.
21
RAID-DP vs Other RAID Methods
RAID Level
Failure Protection
# of Disks
(Cost Hit)
Performance Impact
Multiple disks,
except mirror pairs
2N
2x writes; none to add or
(N or 100%) reconstruct
1 failed disk or
uncorrectable disk
(hard) errors
XOR calculation on writes, and
N+1; N7
when disks added or data
(1/N or 14%)
reconstructed
NetApp
RAID 4
1 failed disk or
uncorrectable disk
(hard) errors
XOR calculation on writes and
N+1; N7
when data reconstructed;
(1/N or 14%)
none when adding disks
NetApp
RAID-DP
2 failed disks or
1 failed disk and
hard disk errors
XOR calculation on writes and
N+2; N14
when data reconstructed;
(2/N or 14%)
none when adding disks
RAID 1
(Mirroring)
RAID 5
(“RAID”)
RAID-DP provides RAID 1 protection…
…with RAID 4 cost, performance and simplicity
© 2008 NetApp. All rights reserved.
22
Protect Oracle Data with SnapLock
Tablespaces
Datafiles
Fabric Attached Storage (FAS)
(partitions)
Q1
Q2
High
Performance
High Perf.
(SnapLock)
2005
2004
Oracle
Database
SnapLock software
helps ensure the
permanence,
accuracy, integrity,
and security of data
by enabling business
records and Oracle
data files to be both
unalterable and
rapidly accessible
online for long
periods of time.
Low Cost
ATA Drives
Online Archive
(SnapLock)
2003
© 2008 NetApp. All rights reserved.
2002
2001
/oradata/2001a.dbf
Unified
Storage
Architecture
/oradata/2001b.dbf
23
Migrate Data Across Storage Tiers
 Periodically move data between
storage tiers as access patterns
change
High Performance
Storage Tier
Low Cost
Storage Tier
Partition
Rules
– e.g. MOVE PARTITION holding Q2
Orders from high performance storage
tier to low cost storage tier
Move important data on demand
– UPDATE of partition key will cause
row to move to a new partition
 e.g. product warranty expires
Use Cases:
Content Mgmt., ILM,
Compliance, Litigation,
Etc.
© 2008 NetApp. All rights reserved.
 Write protect data with SnapLock on
any storage tier
– e.g. MOVE PARTITION to SnapLock
volume and mark as “read only” and
set the retention date.
24
Protect Against Oracle Block Corruption
Oracle® Database
Prevention is better than
recovery…
eBay suffers three outages in three days
Database
Block Write
“… the larger outage was
caused by a corruption of data
files. The corrupted files
began to replicate and quickly
spread across the system, …”
April 11, 2002
Volume Manager
I/O Driver
SAN / NAS
Cloud
Operating System
Host Bus Adapter
Hardware and/or
firmware failures,
misconfiguration,
etc.
Data passes through many layers before reaching the storage subsystem.
Corruption can occur at any point in the data path.
© 2008 NetApp. All rights reserved.
25
SnapValidator: Prevents Oracle block
corruption
 SnapValidator™ is NetApp’s implementation of Oracle’s
HARD initiative
– HARD stands for Hardware Assisted Resilient Data
 Validates writes to volumes that contain Oracle data
– Oracle checksum verification
 Feature of Data ONTAP
 Supported with Oracle Database 9i and 10g (NFS,
iSCSI, FCP)
 NetApp is the only HARD partner to support this
initiative across its entire product line.
 NetApp is the only HARD partner to support
filesystems.
© 2008 NetApp. All rights reserved.
26
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
27
Disaster Recovery Planning
Primary Data Center
Secondary Data Center
Database
Server
Database
Server
Dedicated
Lines and
Switches
RAID0 + 1
Mirror Copy
RAID0 + 1
Mirror Copy
© 2008 NetApp. All rights reserved.
Challenges
 4x actual storage requirement
 Expensive transport
infrastructure
 Cumbersome to install and
manage
 Difficult to keep backup copies
synchronized
28
Disaster Recovery with SnapMirror®
Production
Site
Remote Disaster
Recovery Site
SnapMirror
© 2008 NetApp. All rights reserved.
 Fast data replication and
failover
 Significant reduction in
storage required
 Sync or async replication
 More efficient network
utilization
 Access to mirrored data
 Volume or qtree replication
 Secondary can become
primary in case of disaster at
primary site
 Easy to install and manage
over existing IP infrastructure
(WAN)
29
Unified and Tiered Disaster Recovery with
ReplicatorX®
SAN
SAN / iSAN
Current Instance
Or any
Legacy
Storage
HA/BC site
FAS Storage
FlexClone
Point-in-Time Clones
 Data replication and recovery over any distance in a
heterogeneous environment
 100% data consistency across any number of servers,
storage devices, and fabric-based devices
 Guarantees block write order
© 2008 NetApp. All rights reserved.
30
NetApp’s Value to the DBA
 Performance
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
31
Database Cloning and the Application
Development Process
 Full or partial database
copies required for:
PROD
DEV MAINT
SECONDARY (DR)
TEST/QA
RPT/ETL
© 2008 NetApp. All rights reserved.
– App and DB Development
– Maintenance (OS, DB
upgrade)
– Test and QA
– Training and Demos
– Reporting and DW ETL
 Ability to do this quickly,
correctly, and efficiently
directly impacts Application
Development and
Deployment
32
Application Development and Testing
Example: Traditional Approach
Production
Mirrored Copy
Dev 1
Dev 2
Production database
Mirror copy
Development copies
Testing copies
Total:
100GB
100GB
300GB
300GB
800GB
Dev 3
 8x actual storage requirement
 Time consuming
 Resource overhead
Test 1
Test 2
© 2008 NetApp. All rights reserved.
Test 3
33
Application Development and Testing
Example: NetApp Approach
Production
Mirrored Copy
Dev 1
Test 1
Dev 2
Test 2
Production database
Mirror copy
Development copies
Testing copies
Total:
100GB
100GB
30GB
30GB
260GB
Dev 3
Test 3
 Over 67% reduction in storage
required
 Near instantaneous copies
 Negligible overhead
 Ability to have many more test and
dev copies
more clones = higher productivity
Assumption: up to 10% change in data in the test and dev environments
© 2008 NetApp. All rights reserved.
34
Database Maintenance with FlexClones
Production
Mirrored Copy
Benefits
 Instantaneous copies
 Low resource overhead
Dev 1
Production
DB Clones
Test 1
Snapshot
Dev 2
Test 2
Active
Filesys or
LUN
A B
Dev N
Test N
FlexClone
C D D’ Y
© 2008 NetApp. All rights reserved.
 Inexpensive
 Easy and relatively inexpensive
way to make copies of a production
database without impacting the
database
 Use database clones to test
migrations, apply bug fixes,
upgrades, and patches
 Easily reallocate storage for a
database and its transaction log
files
35
Cloning for Production Support Applications
EMC to NetApp example using ReplicatorX
Production Environment
Production Support Environment
ReplicatorX
ReplicatorX
Continuous Replication
• Development and Test
• Reporting
• Analytics / Data Warehouse
SAN / iSAN
SAN
Current Instance
Or any
Legacy
Storage
Instantaneous Clone
FAS Storage
FlexClone
Point-in-Time Clones

Continuous replication with no disruption to production

Avoid duplicating expensive storage for production support

Local or remote environments

Integrated, instantaneous cloning of current production data

Manage up to 255 clones with minimal storage capacity
© 2008 NetApp. All rights reserved.
36
New Database Development Methodology
 Mirror PROD for initial copy (DR)
PROD
Secondary (DR)
Clones
– If non NetApp PROD, use
ReplicatorX for SAN-to-SAN
 Clone database replicas as
needed
 Create Snapshot™ copies of
replicas for instant
SnapRestore® of working
databases
 Exploit multiple mirrors to
increase fan-out and support
multiple sites
Develop ● Test ● Deploy
© 2008 NetApp. All rights reserved.
37
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
38
NetApp Adds Value to Oracle ASM
Oracle ASM
NetApp
Oracle ASM + NetApp
Protect against Single Disk Failure
Yes
Yes
Yes
Protect against Double Disk failure
No
Yes
Yes
Passive Block corruption detection
Yes
Yes
Yes
Active Block corruption detection
Yes
Yes
Yes
Lost disk write detection
No
Yes
Yes
Stripe data across ASM Disks
Yes
No
Yes
Balance I/O across ASM Disks
Yes
No
Yes
Stripe data across Physical Disks
No
Yes
Yes
Balance I/O across Physical Disks
No
Yes
Yes
I/O prioritization
No
Yes
Yes
Free space management across
physical disks
No
Yes
Yes
Thin provisioning of ASM Disks
No
Yes
Yes
Space efficient Cloning
No
Yes
Yes
Storage Snapshot based Backups
No
Yes
Yes
Storage Snapshot based Restores
No
Yes
Yes
Data Resilience
Performance
Storage Utilization
Data Protection
© 2008 NetApp. All rights reserved.
39
NetApp’s Value to the DBA
 Performance and Utilization
 Backup and Recovery
 Data Protection
 Disaster Recovery and Business Continuance
 Test / Development
 ASM
 SnapManager for Oracle
© 2008 NetApp. All rights reserved.
40
SnapManager® for Oracle®
SnapManager (GUI)
Oracle
Databases
SnapDrive™
NFS, FCP, or iSCSI
NetApp Appliance
NetApp Storage
Appliance
© 2008 NetApp. All rights reserved.
 Automated, fast, and
efficient
 Uptime AND
performance
 Simplify backup,
restore, and cloning
 Tight Oracle
Database 10g
integration
– Automated Storage
Manager (ASM)
– RMAN
41
Oracle Advanced
© 2008 NetApp. All rights reserved.
42
Other Cool Stuff…
 NFS,v4
– Support with Data ONTAP 7.3
 Planned
– Feature implementation with DNFS for 11gR2
 Workload prioritization
 Smart redirection
 RAC Failover Management with NetApp
– Integrate Manage ONTAP API’s into CRS…
 If Storage not accessible, is it failing over? (planned or
unplanned)
 Alternative: node eviction + “suicide” + application/user must
reconnect…
 Oracle Enterprise Manager Grid Control
– Enhanced plugin (to include blocks)
– Connector (to Operations Manager)
 SMB/E Solution with DNFS+SE-RAC on Windows…
© 2008 NetApp. All rights reserved.
43
FlexShare Key Features
 Relative priority of
different volumes
 Application (user) Vs.
Data ONTAP
background (system)
prioritization
Without FlexShare™
High Priority
Volumes
Medium Priority
Volumes
Low Priority
Volumes
0
20
30
40 50 60
Latency (msec)
70
80
10
20
30
40 50 60
Latency (msec)
70
80
With FlexShare
High Priority
Volumes
Medium Priority
Volumes
Low Priority
Volumes
0
 Per-volume cache
policies
Prioritize Client over System
Prioritize System over Client
System Load (IOPS)
 Dynamic configuration
updates
10
System
Client
Time
© 2008 NetApp. All rights reserved.
44
FlexShare High Benefit Use Cases
 Consolidated Environments
– Production, Test, and Development on the same system
 Backup/Disaster Recovery Throttling
 Manage “like” data with consistent performance
– Structured and unstructured
 Match NetApp cache management with Oracle cache management
© 2008 NetApp. All rights reserved.
45
Oracle Enterprise Manager Grid Control
Plugin for NetApp
Monitor Trends and Threshold Alerts
Monitor Key Statistics
Monitor Utilization
Etc.
•Ships with Oracle
Enterprise Manager
•Developed,
maintained and
licensed separately
by Oracle
© 2008 NetApp. All rights reserved.
46
Introducing Database NFS - DNFS
 What is DNFS?
– An Oracle developed and supported NFS client for
all platforms. (Including Windows)
 What are the benefits of DNFS?
– Simpler management with enhanced performance
and scalability.
– No certification required  immediate customer
availability
 Fully tested during development rather than after
deployment
– Supports RAC, ASM, and HARD across all
operating environments
© 2008 NetApp. All rights reserved.
47
Performance
 Automatic Link Aggregation
– Users can specify multiple ethernet connections between the
Oracle database and the NFS server
– DNFS will distribute the load over all active links in a
balanced fashion
 Improved I/O Concurrency
– Supports fully concurrent, asynchronous I/O on all platforms
 Summary
– Any Oracle application can easily take advantage of the
performance available on multiple parallel wires
– No other NFS client can do this…
High performance and high throughput, regardless of
operating environment and without the need for expert tuning
© 2008 NetApp. All rights reserved.
48
Simplicity
 Runs outside of the OS kernel
– No integration with operating system’s virtual
memory or buffer cache
 DNFS has a shorter code path  faster than other
NFS clients with fewer bugs
 No mount options
– Already tuned for database operations
– Simple to set up  fewer operational errors
– Fewer support calls
© 2008 NetApp. All rights reserved.
49
Database Backup with A-SIS Deduplication
(Advanced Single Instance Storage)
No perceivable performance impact
Remote Office
–Write Overhead ~1%
–Read Overhead 0%
Fortune 500 Manufacturer
Data Center
Backup
Server
• Backing up Oracle Databases to
NearStore
A-SIS
deduplication
NetApp
Storage
50:1
Space
Savings
Backup Data
• Reported 50:1 space savings with
A-SIS deduplication vs tape
backups
• Backups are quickly recovered
from disk
• Initial database backup is
deduplicated and reduced in size
Heterogeneous
Storage
NearStore
Database
Servers
© 2008 NetApp. All rights reserved.
• Each subsequent backup
provides space savings through
deduplication
50
Question and Answers
© 2008 NetApp. All rights reserved.
51