QLogic 10000 Series Accelerating Database Infrastructure Using Oracle Real Application Clusters 11g R2 and QLogic FabricCache© Adapters - Manual (Page 2)

Accelerating Database Infrastructure
Using Oracle Real Application Clusters 11gR2 and QLogic FabricCache
1
Executive Overview
Thousands of companies world-wide use Oracle Real
Application Clusters (RAC) running on cost effective server
hardware, to run mission-critical applications for their business.
Today’s more powerful servers, increased use of virtualization
and increased datacenter density allow for large-scale
implementations of database applications based on Oracle RAC.
This presents challenges for shared storage subsystems to keep
up with demands for highly-accessible, highly available shared
storage for database systems such as Oracle RAC.
Storage subsystems, consisting of large arrays of spinning disks, have been challenged to provide
adequate I/O capacity and transaction latency to support applications such as Oracle RAC that require
high performance. While using flash memory for server-side I/O acceleration has become a popular
method to bridge this performance gap, clustered applications such as Oracle RAC have an additional
requirement for shared access across all nodes that limits the viability of typical "Captive Cache"
solutions. A new "Shared Cache" paradigm is needed to support clustered application environments to
allow maximum utilization of datacenter server resources.
Introduction
This paper introduces the concept of database infrastructure acceleration through the use of caching
host bus adapters (HBAs) to optimize server storage throughput across all nodes within the cluster.
The enabling technology for Oracle RAC database infrastructure acceleration is provided by QLogic in
the form of their FabricCache™ adapter technology. The benefits of using FabricCache adapters for
RAC database storage access are discussed here, as are topics such as implementation considerations
and costs.
What is Database Infrastructure Acceleration?
Over the past decade, the cost of server hardware used for running enterprise databases has gone
down dramatically. However, to meet business demand, IT infrastructures have also grown
dramatically over the same time span. This growth presents challenges to IT: unevenness in business
demand, added application functionality, difficulty scaling management and support, and managing
cost. In order to be a strategic asset to the business, IT needs to face such challenges proactively.
A large amount of IT assets are typically used to manage databases, which are in turn strategic assets
for the business. For IT infrastructure used to manage databases to be “optimized”, that infrastructure
should be simple to manage, effectively utilized, protected from failure, and able to deliver maximum
value to the business at all times.
A key factor in delivering an optimized database infrastructure is providing adequate resources for
storage I/O. Database management systems generate large amounts of I/O, and the performance they