The Ultimate Scale-Out Platform for Oracle 10g
Delivered by:
Matthew Zito, GridApp
John McAbel, IBM
156 5th Avenue
Penthouse
New York, NY 10010
P: 646.452.4100
www.gridapp.com
[email protected]
GridApp and IBM Welcomes You!
• Thank you for attending
• Our presentation will last approximately 40
minutes with a live Q&A to follow
– Topic: The Ultimate Scale-Out Platform for Oracle 10g
– Today’s Presenters:
• John McAbel, Worldwide Product Marketing Manager,
IBM System x
• Matthew Zito, Chief Scientist, GridApp Systems
[email protected]
IBM System x™
IBM BladeCenter
The Ultimate Scale-Out Solution
for Oracle Database 10g
John McAbel
WW Solutions Manager
[email protected]
| Jan 2007 |
© 2007 IBM Corporation
IBM System x
Key Challenges for Oracle Database Managers
 Pressure to upgrade from Oracle8i or Oracle9i to
Oracle EBS version 12 only supports Oracle 10g
How easy is it to map settings from 8i or 9i to 10g?
What is the additional cost to manage a RAC environment?
 Need to lower IT total cost of ownership
Under-utilization of servers
Cost of maintaining present operating system license fee structure
Is Linux ready to handle the complexities of large data centers?
Is each new server a capital budget item?
 Ability to meet Service Level Agreements (SLA’s)
Increasingly difficult as IT infrastructure ages
Database administrator skill decline with employee turnover
 Hardware vendors forcing “a one size fits all” concept
UNIX-based SMP vs. “Rack and Stack”
 Oracle performance proof points for the proposed solution?
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
The IBM & Oracle Relationship
 IBM and Oracle a strong Relationship for 19+ Years
(27 years with JD Edwards, 16 with PeopleSoft)
 Oracle managed as an Integrated Account
 Enhanced a Strong BCS Relationship –
5000 Skilled consultants
 IBM Viewed by Oracle as a Significant Ally
Jeff Henley, Chairman, is IBM Executive Sponsor
Charles Phillips, President, has IBM alliance responsibilities
 Strong Technology Relationship
 Technical Sales Education Sessions
 IBM participation in Oracle Customer Events
 Shared “pay as you grow” Strategy
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Blade Market Share Overview
IBM has been #1 for 13 consecutive quarters!
60%
50%
41%
40%
34%
30%
20%
12%
10%
Q
30
3
Q
40
3
Q
10
4
Q
20
4
Q
30
4
Q
40
4
Q
10
5
Q
20
5
Q
30
5
Q
40
5
Q
10
6
Q
20
6
Q
30
6
0%
IBM
HP
Dell
Others
Source: IDC Quarterly Server Tracker, 11/2006
Total Blade Server Market
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Shared “pay as you grow” Vision
 Oracle and IBM BladeCenter offer economical, upgradeable products
that allow you to make a small initial investment then grow as the needs
of the business grows
IBM’s Enterprise X-Architecture delivers servers based on modular, scalable designs
Maintain Oracle Database 10g™ investment through seamless upgrade from Standard to
Enterprise for increased levels of system availability
1
0
g
O
r
a
c
l
e
D
a
t
a
b
a
s
e
IBM BladeCenter
P
e
r
f
o
r
m
a
n
c
e
Performance
S
y
s
t
e
m
P
r
o
v
i
s
i
o
n
i
n
g
W
o
r
k
l
o
a
d
m
a
n
a
g
e
m
e
n
t
S
c
a
l
a
b
i
l
i
t
y
A
u
t
o
m
a
t
e
d
S
t
o
r
a
g
e
M
a
n
a
g
e
m
e
n
t
I
n
t
e
g
r
a
t
e
d
C
l
u
s
t
e
r
w
a
r
e
A
v
a
i
l
a
b
i
l
i
t
y
E
n
d
t
o
e
n
d
T
r
a
c
i
n
g
P
r
e
d
i
c
t
i
v
e
B
e
h
a
v
i
o
r
R
e
p
o
r
t
s
S
y
s
t
e
m
H
e
a
l
t
h
C
h
e
c
k
s
B
e
s
t
P
r
a
c
t
i
c
e
A
d
v
i
s
o
r
i
e
s
S
y
s
t
e
m
s
M
a
n
a
g
e
m
e
n
t
D
e
p
l
o
y
m
e
n
t
A
u
t
o
m
a
t
i
o
n
C
e
n
t
r
a
l
i
z
e
d
M
a
n
a
g
e
m
e
n
t
C
o
n
s
o
l
e
Intel®, AMD®, or POWER® Processors
Enterprise X-Architecture™
Open Design
Scalability
Add 2-way or 4-way Blades
FC and Network Switches
Capacity On Demand
Availability
Predictive Failure Analysis
OnForever™ Initiative
Dual GB Ethernet Backplane
Calabrated Vector Cooling™
Light Path Diagnostics™
Systems Management
IBM Director
Integrated Systems Management
Processor
* See www.ibm.com for specifics/server
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Nodes Deployed per RAC Cluster (Linux)
Source: Erik Peterson, Oracle RAC Development, Oracle Open World, October 2006

85% of Linux deployments are 4-nodes or less

Horizontal scale-out often required for Linux AMD/Intel based architectures
In many cases, involves 2-CPU or 4-CPU nodes

Some other architectures offer scale-up as well as scale-out alternatives
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
IBM BladeCenter One Family
Investment Protection
New IBM
BladeCenter
HT
BladeCenter T
BladeCenter H
BladeCenter
Announced: Feb. 2006
Announced: Nov. 2002
Announced: Apr. 2004
14 Blades, 7U
8 Blades, 8U
Ruggedized Chassis
Enterprise & SMB Chassis
Telco, Military,
Medical Imaging Apps
Mainstream Applications
Remote Sites (stores)
14 Blades, 9U
High Speed (>10GB)
Extreme I/O for data
intensive environments
Common Blades, Common Switches
Full performance and manageability of rack-optimized platforms ...
... at TWICE the density of most comparable non-blade 1U servers
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Blade Portfolio Continues to Build...
NEW!
IBM HS21
AMD Opteron LS21
AMD Opteron LS41
 Two socket AMD
 Four socket AMD
 Single/Dual core
 Single/Dual core
 32-bit/64-bit
solution for Linux &
 4 GB Modular Flash
AIX 5L™
Drive - NEW
 Performance for
deep computing
clusters
 32GB Memory
 64GB Memory
 Similar feature
set to HS21
 Start with two
sockets then
grow to four
 Edge and mid-tier
workloads
 32- or 64-bit HPC,
VMX acceleration
 Collaboration
 UNIX server
consolidation
 High memory
bandwidth
apps
 High memory
bandwidth
apps
 Intel Xeon DP
Target Apps
Features
 Single, Dual, Quad
IBM JS21
 Two PowerPC®
processors
 32GB Memory
 Web serving
Quad
Core
Common Chassis and Infrastructure
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
BladeCenter Energy Efficiency
TCO: Lowering Power and Cooling Design Saves Money
HP Power Usage vs IBM Blades
22%
36%
53%
Industry
1U

Intelbased
Blades
AMDbased
Blades
Industry 1U
HP Blade
IBM Blade
IBM HS20
w/ ULP
Saving power is key to lowering Total Cost of Ownership
Less Power
Less Heat
Less A/C
Source:Jan
HP2007
and IBM
Power Tools
IBM BladeCenter for Oracle
|
Less Racks
Less Space
© 2007 IBM Corporation
IBM System x
Typical Cluster
Configuration
Step 1
Integrate Servers
Public
Internet/
Intranet
Clients
Firewalls
Layer 4-7
Switches
SSL Appliances
Layer 2
Switches
Caching Appliances
- Intel 32-bit based
Routers
(Layer 3
Switches)
SSL Appliances
Caching Appliances
- Intel 32 bit based
Eight Database Servers
- POWER based
Storage
Fibre
Switches
Storage
Fibre
Switches
Storage
Area
Network
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Result
IBM BladeCenter
Collapses Complexity
Public
Internet/
Intranet
Clients
Firewalls
Routers
(Layer 3
Switches)
Storage
Area
Network
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Simplifying Datacenter Topology
9
5
2
7
3
6
10
8
4
1
Typical Datacenter Configuration
1.
2.
3.
4.
5.
Ten x86 1U 2-way servers
RISC-based 2-way server
HPQ 4-way server
Alteon L7 E’net switches
FC SAN switches / Cables
6.
7.
8.
9.
10.
IBM BladeCenter for Oracle
Layer 2 GbE switches
KVM switches
Ethernet cables
KVM cables
Power cables
Jan 2007 |
Bladed Datacenter Configuration
IBM BladeCenter
© 2007 IBM Corporation
IBM System x
BladeCenter Chassis (back view)
Gb Ethernet
Module
Power Supply
Module
Blower
Module
IBM BladeCenter for Oracle
Management
Module
Jan 2007 |
© 2007 IBM Corporation
IBM System x
An Open IBM BladeCenter Ecosystem
 Open Specification – Released September 2004 by IBM and Intel
260+ companies received the BladeCenter specs
“By opening up the
specification and
making it royalty-free, it makes it very easy
to do business and create products
around the platform.”
“This
will grow the
market. It will be
better for mainstream customers, smaller
customers, and customers with specific
needs.”
-John Humphreys, IDC
-Krish Ramakrishnan, Topspin




BladeCenter Alliance Program – Over 350 partners
IBM BladeCenter Partner Solutions Interoperability Lab
IBM Engineering and Technology Services BladeCenter Open Support Center
Blade.org
Founding members include: Brocade, Cisco, Citrix Systems, Intel, Network Appliance,
Nortel, Novell, VMware.
Oracle has agreed to provide input and guidance to shape future direction
A collaborative organization focused on accelerating the expansion of solutions for
BladeCenter
Intended to assist solution providers in developing applications and extending BladeCenter
into vertical industry
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
IBM BladeCenter Increasing availability on three fronts…
Failure Masking
Failure Avoidance
Minimized Outages
Chipkill Memory
Predictive Failure Analysis :
HDD, VRMs, Memory,
CPU, Fans, Power Supplies
Hot-plug components
Dual Ethernet Backplane
IBM Director
Light-Path Diagnostics™
Hardware
Redundancies
Built-in temperature and
voltage monitor systems
Clustering
Dual hard disk
per blade
Integrated System
Management Processor
Calibrated
Vectored Cooling™
Continuous Computing for Database Solutions
Keep it running: Increasing mean time between failures (MTBF)
Faster fix times: Decrease mean time to repair (MTTR)
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
IBM BladeCenter H Oracle 10g FDC with AMD® Blades
AMD-based
FDC Analysis
Oracle 10g
IBM BladeCenter H
(4) LS20 Blades
Red Hat EL 4,
Oracle Database 10g
IBM TotalStorage DS4500
(6) EXP700 Storage Expan. Units
(84) 36.4GB HDD
Two Oracle Databases utilized:
OLTP = Order Entry (Size 1.1 TB)
DSS = performed queries on credit
Figure 1. Components of the Flexible Database Cluster System
The Primary Goals for this new proof of concept were:
 The IBM BladeCenter H architecture and technology provide a high-availability platform for
the Flexible Database Cluster.
 Capacity can be increased dynamically and transparently, without user interruption, to
reduce workload completion time.
 Scalability is directly related to I/O throughput. With the IBM TotalStorage SAN
architecture, adding disk drives to the array may resolve performance bottlenecks.
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Time to Complete (min)
IBM BladeCenter H Oracle 10g FDC Results
Decision Support System (DSS) Workload
Reduction in Completion Time
70
60
50
40
30
20
10
0
1 Node
2 Nodes
Blades
4 Nodes
Results:
• Oracle 10g RAC scales on demand
• Sorted table had 1.6 billion rows
• Sorting not performed entirely in memory
• Adding nodes a non-intrusive effort
• No scalability limits seen at software level
• 1 node = 59 min (564,972 row/sec.)
• 2 nodes = 31 min (1,075,269 row/sec.)
• 4 nodes = 18 min (1,851,852 row/sec.)
• Scalability was 82% from 1 to 4 nodes
Transactions per Second
On-Line Transaction Processing (OLTP) Workload
1200
Results:
1000
• Scalability directly related to I/O throughput
• No server scalability problems experienced
• Near-linear scaling 87% through 4 nodes
• 1 node (310 TPS) to 4 nodes (1081 TPS)
800
600
400
200
0
1 Node
2 Nodes
3 Nodes
4 Nodes
Blades
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
Based in New York, NY, The Street.com is a leading provider of financial services
information. Consumers use the internet to access information and product services.
Business Need
Desire to upgrade to Oracle 10g with Real Application Clusters
& purchase additional licenses
Highly flexible database architecture to better manage the
constantly changing demands of their business.
Required fully automated turnkey solution that ensured
maximum uptime, performance and scalability.
Solution
GridApp D2500 Database Appliance Solution
Three IBM BladeCenter Chassis each with:
10 dual core AMD blades, Red Hat Enterprise Linux 4
2 Cisco switches, Optical Pass Thru Module
Oracle Database 10g RAC licenses
Why IBM BladeCenter Selected
Proven price/performance with Oracle 10g RAC
“We selected IBM BladeCenter
to run Oracle 10g RAC and
GridApp software due to its
superior design. AMD-based
blades were the clear choice
after considering their
scalability, value and proven
price/performance ratio.”
IBM Blades make scaling simpler and efficient.
Value of AMD Opteron Processors
Easily scale out and not have to linearly scale up was an
attractive economic alternative
Alex Spinelli
CIO
The Street.com
GridApp’s industry only next generation database platform
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
IBM System x
IBM BladeCenter – It’s All About Choice...Choose IBM
 Leading technology wrapped in openness
#1 Market Share Leader (42% IDC 2006)
Four Chassis and Four blades to choose from
Intel, AMD, POWER processor
Chose your OS: Linux, Windows, AIX, Solaris 10
Help cut your cluster cabling by 75%
IBM PowerExecutive – Monitor real power usage
Novell SLES Chassis-wide License
Vast Networking Options to chose from – Opens Specs
Built in availability features
Blade.org
Many proof of concepts & Happy customers!
 Global support coverage through six IBM/Oracle
International Competency centers
USA (3), France, Canada, Japan
Delivering
A
 IBM’s commitment to Linux
Complete
Linux Technology Center
Infrastructure
 IBM Global Services
Solution Support
Professional Services
Solution Development
(Testing/Certification)
Oracle
Application
Cluster File System
Hi Av - Failover
Oracle Database
Operating System
Hardware
Largest Oracle Practice (Outside of Oracle)
IBM BladeCenter for Oracle
Jan 2007 |
© 2007 IBM Corporation
Scaling out IBM Blades & Oracle with
the Next Generation D2500
Delivered by:
Matthew Zito, Chief Scientist
156 5th Avenue
Penthouse
New York, NY 10010
P: 646.452.4100
www.gridapp.com
[email protected]
Clustering Technologies
• Shared-nothing – Each cluster node has
no common resource with any other
node
• Shared-everything – Each cluster node
shares a universal pool of common
resources
• Shared-something – Each cluster node
shares some resources, but not others
• Active-Passive – Only one cluster node
at a time is providing any given service
[email protected]
Oracle RAC
• Oracle RAC is a “shared-something”
cluster
• RAC nodes share:
– Application state awareness (Cache Fusion)
– A single set of on-disk data (ASM, OCFS,
etc.)
• RAC provides
– Enhanced reliability
– Enhanced scalability
– Reduced processing cost
[email protected]
What RAC Looks Like
[email protected]
Processing is Getting Cheaper…
60
US$ per TPC Run
50
40
30
20
10
0
3/15/2000
7/28/2001
12/10/2002
4/23/2004
9/5/2005
1/18/2007
Date
[email protected]
…and Servers are Getting Smaller
• 8/23/06 – IDC Server Shipments
– Enterprise Servers (>$500k) – down 6.9%
– Midrange Servers ($25k-$499k) – down 3.5%
– Volume Servers (<$25k) – up 6.2%
• Organizations are dramatically
– Decreasing the size of the average deployed
server
– Increasing the number of total servers
[email protected]
Incremental Scalability
• Buy only what you need today
• Take advantage of the latest technology
– Moore’s law works for you, since you can
buy faster servers as released and add
them to the cluster
– Incrementally swap out older, slower
hardware for faster hardware
• GridApp D2500 allows you to scale
clusters online in minutes
[email protected]
GridApp’s D2500 Database
Appliance
• Enterprise-grade database appliance for easy
administration and fast deployment of Oracle
RAC
• Turnkey virtualized and distributed database
solution in a blade form factor
• Start with as few as 2 processors, grow to
hundreds
• Big box/enterprise manageability at a costeffective price point
[email protected]
The D2500 – Next Generation
Database Platform
• Best-of-Breed Reference Architecture
– Time-tested RAC platform
– Validated infrastructure
• Over 100 Deployments
• Automation to Simplify the Deployment of RAC
– RAC in a day
– One-click scalability
• GridApp’s Powershare Technology for Dynamic
Resource Allocation
– Performance When and Where You Need It
• Deploy a highly-available RAC environment in
a day
[email protected]
How Does It Work?
• Hardware - IBM BladeCenter
• Processors – AMD or Intel Processors
• Operating System - Red Hat Enterprise
Linux
• Database - Oracle 10gR2 RAC
• Management Software - GridApp Clarity™
The D2500 is an all-in-one solution for your
database environment
[email protected]
GridApp Clarity™
• Automates the deployment of Oracle RAC
– Average time to deploy RAC manually - 1-2 weeks
– Average time to deploy RAC with Clarity - 45 minutes
• Adds additional HA to Oracle RAC
– “Hot Spare” nodes
– Helps database sustain failures across high-transaction
time periods
• Integrates with the hardware and OS
– A single picture into your database environment
– Integrated alerting
• One-click Scalability of Oracle RAC
[email protected]
GridApp Clarity™
[email protected]
Customer Case Study
• High-throughput transaction
processing database
• Successfully scales from 4 to 10
nodes on-demand to accommodate
load
• Three sites - production,
development, disaster recovery
Cost savings: >$1m
[email protected]
Summary
• Oracle RAC is the next step in
database infrastructure
• GridApp’s D2500 database
appliance is the next generation
database platform for RAC
• GridApp and IBM together are the
optimum scale-out platform for
RAC
[email protected]
My Contact Details
Email: [email protected]
[email protected]
Schedule a Demo
• Email [email protected]
• SUBJECT LINE = Demo
[email protected]
Get the Companion
D2500 White Paper
• Email [email protected]
• Subject Line = D2500 White Paper
• http://gridapp.com/resources/whitepaper
_gateway.php
[email protected]
Q&A
with
Matthew Zito and John McAbel
[email protected]
Thank You!
[email protected]