3. The Traditional Data Center
• Dedicated silos are inefficient
• Sized for peak load
Middleware
• Constrained performance
Database
• Difficult to scale
• Expensive to manage
Storage
Dedicated Stacks
6. Grid Economics for the Data Center
Virtualization and clustering Lower CapEx & OpEx
enable consolidation
Pay-as-you-go scale-out Avoid upfront CapEx & OpEx
High Quality of Service Avoid lost user productivity,
improve customer service
Automated grid management Raise IT staff efficiency,
lower OpEx
7. Consolidation with Grid Computing
Application A Application B Application C Application D Application E
Workload Avg Utilization
<20%
• Take advantage of
complementary
workload peaks
Server A Server B Server C Server D Server E • Higher utilization rates
and efficiency
Virtualization and clustering enable consolidation
• Lower CapEx & OpEx
Oracle Shared Instance
Applications A, B, C, D, E
• Green footprint
Net Avg Utilization
Workload 70%
Freed capacity to
deploy elsewhere
Server A Server B Server C Server D Server E
8. Scale Out with Grid Computing
Oracle Shared Instance
Applications A, B, C, D, E • Pay-as-you-go scale-out
• Lower upfront CapEx and
Net
Workload
If utilization too high,
increase capacity
ongoing OpEx
• Green footprint
• Rightsized capacity planning
Server A Server B Server C Server D
• Smaller, standard machines
Scale-out on-demand running at higher utilization
• World-class clustering at all levels: • Defer equipment procurement
database, middleware, storage • Exploit advances in hardware
• Add/Remove nodes on-demand price-performance and energy
efficiency
• Scale out as workload increases
9. Quality of Service with Grid Computing
Oracle Shared Instance
Applications A, B, C, D, E • Systematic high Quality of
Service
Net
Workload
• Reliability through redundancy
Server A Server B Server C Server D Server E
• Predictable performance at
any scale
High performance and availability
• High availability – every
• Load balancing • Disaster recovery application gets HA
• Failover • Rolling upgrades
• Active-Active operation
10. Efficient Management with Grid Computing
• Deploy standard virtual machine
images quickly and easily
• Expedited provisioning and patching
• Manage Quality of Service from end-
user perspective
• Automated diagnostics and tuning
• Real-time and predictive monitoring
• Comprehensive testing and validation
11. Most Complete Grid Stack in the Industry
Grid Computing in All Tiers
Middleware
• Application Grid
• WebLogic Server
• Coherence In-Memory Data Grid
• JRockit Real Time
• Tuxedo
Database
• Real Application Clusters
• In-Memory Database Cache
• HP Oracle Database Machine
Storage
• Automatic Storage Management
• Oracle Advanced Compression
• Exadata Storage Server
Infrastructure
• Oracle VM
• Oracle Enterprise Linux
Management
• Oracle Enterprise Manager
12. Oracle Real Application Clusters
Beats SMP Premium, Pay as you Grow
Year 1 Year 2 Year 3 Year 4 Year 5
Large SMP
High Upfront Cost High Ongoing Maintenance Costs Diminishing Returns
RAC Cluster + +
Low Cost Building Blocks Lowest TCO Greener and More Powerful
13. Case Study: Mercado Libre
eBay of Latin America
Mercado Libre
Number RAC Nodes vs Workload
17. Linux and Virtualization
Lowest Cost Platform
Oracle Enterprise Linux
Unbreakable Linux Support
Dedicated engineering team
Free Oracle Management pack for Linux
Faster Deployment
Oracle Validated Configurations
for Linux and Oracle VM
Oracle VM Oracle VM templates
24X7 support, 145 countries, 27 languages
Free to download, use and distribute
Real-world testing and validation
18. RAC One Node
Better Virtualization for Databases
• A virtualized single instance
database
• Delivers value of server
virtualization to databases on
physical servers
• Live migration of instances
across servers
• Rolling patches for single
instance databases
• Built-in cluster failover for high
availability
• Online upgrade to RAC
• Standardized deployment
across all Oracle databases
19. RAC One Node Deployment
Server A Server B Server C
DB1 DB2 DB3 DB4 DB5
Common Shared Storage
Single Cluster
20. Omotion
Client Connections
Server A Server B Server C
DB1 DB2 DB3 DB4 DB5
Common Shared Storage
Single Cluster
21. Omotion
Client Connections
Server A Server B Server C
DB1 DB2 DB2 DB3 DB4 DB5
Common Shared Storage
Single Cluster
22. Omotion
Client Connections
Server A Server B Server C
DB1 DB2 DB2 DB3 DB4 DB5
Common Shared Storage
Single Cluster
23. Omotion
Client Connections
Server A Server B Server C
DB1 DB2 DB3 DB4 DB5
Common Shared Storage
Single Cluster
24. Cluster Failover
Server A Server B Server C
DB1 DB2 DB3 DB4
Common Shared Storage
Single Cluster
25. Cluster Failover
Server A Server B Server C
DB1 DB2 DB3 DB4
Common Shared Storage
Single Cluster
26. Cluster Failover
Server A Server B Server C
DB1 DB2 DB3 DB4
Common Shared Storage
Single Cluster
27. Andreas Stephan
Senior DBA Consultant
Bayer Business Service
“We manage thousands of databases and application servers with
Enterprise Manager, and we have been able to reduce the time for
provisioning software from 4 hours down to 1 hour, as well as reduce
patch application time from 1 hour down to 1 minute per database.
Enterprise Manager Grid Control allows us to automate this process,
which translates into huge savings in time and money.”
28. Pressure to Streamline IT Operations
Better Quality
of Service • Many organizations are 30%
below achievable IT productivity
levels Enterprise Management Associates, 2007
Better Agility
Better Agility
Lower Risk
Lower Risk
• 40% of CIOs surveyed cite lack
IT
of automation tools
Operations Enterprise Management Associates, 2007
• 60%–70% of IT budget is spent
on operations and maintenance
Lower CIO Magazine, 2007
Operational Cost
29. Case Study: Oracle IT
Cost Savings using Oracle VM
• 83% reduction in hardware
• CPU utilization increased from 7% to 73%
• Revenue per server increase 5X
• Floor space consumption reduced 50%
• Data center power consumption reduced 40%
• Greatly simplified server refresh 1300
environments automatically provisioned
weekly
• Servers to administrator ratio increased 10X
• SaaS and hosting/management services
• 67% reduction in hardware
• CPU utilization increased from 9% to 55%
30. Grid Economics for the Data Center
Takeaways
Virtualization and clustering Lower CapEx & OpEx
enable consolidation
Pay-as-you-go scale-out Avoid upfront CapEx & OpEx
High Quality of Service Avoid lost user productivity,
improve customer service
Automated grid management Raise IT staff efficiency,
lower OpEx