ISS Technology Update, Volume 7 Number 5 - Newsletter

ISS Technology Update Volume 7, Number 5
13
Implementing a scalable, highly available Oracle® Database for SMBs
Most small to mid-size businesses (SMBs) do not have the resources to deploy highly available servers that are needed for 24x7
access to critical applications. Instead, SMBs usually rely on single server systems, causing them to experience periods of down
time that negatively impact revenue and customer service, and ultimately, customer satisfaction. Even upgrading their server can
cause major disruptions.
Larger corporations with dedicated IT staff and bigger budgets have been able to deploy highly scalable and available
solutions using clustering technologies like Oracle® Real Application Clusters (RAC). Until recently, the cost and the complexity
of setting up this type of infrastructure have limited the ability of SMBs to provide similar levels of service for their customers.
Oracle has made great advances in reducing the complexity and cost of clustering by introducing Oracle® Database Standard
Edition (SE), which includes RAC at no additional charge (for up to four sockets in a cluster). RAC functionality allows customers
to implement a high availability environment at a much lower cost than was previously possible. In addition, the management
of clustered environments has been dramatically improved by complementary Oracle technologies such as clustered file systems
and Oracle® Enterprise Manager. However, the complexity of the hardware configuration and operating system set-up can
still stifle the decision to implement such a cluster environment.
This article describes a scalable and highly available solution using HP BladeSystem hardware with Oracle Database SE
including RAC, which reduces the cost and complexity for SMBs to deploy 24x7 solutions.
Traditional clustering implementations for 24x7 Operations
Prior to the HP/Oracle RAC solution, SMBs had to choose between two traditional clustering options: an active/passive cluster
or an active/active cluster.
Active/passive clusters
Many cluster implementations use an active/passive approach in which the servers and storage are physically connected
together (Figure 5-1), but only one server at a time can run the database and access storage. In the event of a failure, the
database is started on the second server, which requires users to reconnect. This results in downtime and performance issues
that can last several minutes. This may be acceptable in some cases, but there are situations where loss of data access can
have a significant negative impact on business. Examples of active/passive clusters include Microsoft® Cluster Server and HP
Serviceguard. Table 5-1 summarizes the strengths and weaknesses of active/passive clusters.
Figure 5-1. Active/passive cluster example