Scalability and Clustering
363
Titan SiliconServer
High Availability Clusters
The Titan SiliconServer supports two-node Active/Active (A/A) clusters. In A/A cluster
configurations, each server can host independent EVS, which can service network requests
simultaneously. Should either of the servers fail, the EVS from the failed node will automatically
migrate to the remaining server. Network clients will not typically be aware of the failure and will
not experience any loss of service, though the cluster may operate with reduced performance
until the failed server is restored. After the server is restored and is ready for normal operation,
the EVS can be migrated back to the original server.
A license is required to set up an A/A cluster. Contact BlueArc to purchase an A/A cluster
license. For more information on how to enter the license key(s), see "
To Add a License Key
".
Server Farms
A Server Farm is a collection of standalone Titan SiliconServers or HA clusters in which all
storage resources are combined in a shared Storage Pool. Each server or cluster in the Server
Farm can host up to a maximum of 8 EVS. EVS can be easily migrated between the servers and
clusters in the Server Farm through the SMU. There are three primary reasons to configure a
Server Farm:
•
Maximum performance
- When maximum data throughput is required, EVS can be
migrated to a high-end Titan or one on which no other EVS resides. This will ensure that
the EVS enjoys the fully dedicated resources of a single Titan SiliconServer.
•
Load balancing
- Heavily used EVS can be migrated to less-busy Titans, or to higher-
end Titans which support greater server capacity, resulting in more efficient use of
available resources.
•
Failure recovery
- In the event of a catastrophic failure of any standalone server, EVS
hosted by the failed server can be brought online on any other server or cluster in the
Server Farm.
