Databases Reference
In-Depth Information
synchronously to one or more remote locations and greatly widen distances where Data
Guard might be deployed.
In addition to providing physical standby database support, Data Guard can be used to
create a logical standby database where Oracle archive logs are transformed into SQL
transactions and applied to the open standby database. Data Guard also supports snap‐
shot standbys where redo is received but not applied when data is simply to be replicated
to the standby for testing purposes (such standbys can be converted to physical standbys,
and then redo is applied).
If an outage occurs, the Data Recovery Advisor leverages RMAN and Data Guard (in‐
cluding standbys) in determining the best recovery options that minimize any data loss.
Administrators can choose among these options or the Advisor can be set to automat‐
ically run the best choice.
Fail Safe
The Fail Safe feature provides a higher level of reliability for an Oracle Database on a
Windows platform than simple high-availability approaches in leveraging Microsoft
Cluster Services. Failover is implemented through a second system or node that provides
access to data residing on a shared disk when the first system or node fails. Fail Safe is
primarily a disaster recovery tool, so some downtime does occur as part of a failover
operation. The recommended solution for high server availability on all platforms, in‐
cluding Windows, is RAC.
Oracle Real Application Clusters
RAC first appeared as an option for Oracle9
i
, replacing the Oracle Parallel Server (OPS)
option. RAC can provide failover support as well as increased scalability on Unix op‐
erating system variations, Linux, and Windows clusters. Key to RAC's improved scala‐
bility was the introduction of Cache Fusion that greatly minimized the amount of writ‐
ing to disk that was formerly used to control data locks. Oracle Database 10
g
introduced
a new level of RAC portability and Oracle support by providing integrated “clusterware”
for supported RAC platforms.
With Real Application Clusters, you can deploy multiple Oracle instances on multiple
nodes of a clustered solution or in a grid configuration. RAC coordinates traffic among
the systems or nodes, allowing the instances to function as a single database. As a result,
the database has proven to scale across dozens of nodes. Since the cluster provides a
means by which multiple instances can access the same data, the failure of a single
instance will not cause extensive delays while the system recovers. You can simply re‐
direct users to another instance that's still operating. Applications can leverage the Ora‐
cle Call Interface (OCI) to provide failover to a second instance transparently to the
user.
Data Guard can be used to provide automated failover with bounded recovery time in
conjunction with Oracle Real Application Clusters. In addition, it provides client
