Bookshelf v7.5: Clustering Guidelines

Siebel Server Installation Guide for Microsoft Windows > Clustering Your Siebel Deployment for Failover >

Clustering Guidelines

Clustering is a form of high availability (HA) system architecture designed to maximize availability of data and application services for users. Clustering creates a layer of redundancy so that when a function fails, other resources are available to take over the failed function. This process needs to be:

Automatic—no operator intervention should be needed.

Transparent—the users should not need to change anything.

Siebel eBusiness Applications can be deployed using a highly available architecture by employing both load-balancing techniques and failover clustering technologies.

Following the guidelines below will help promote successful failover protection in your system. However, these guidelines are in no way meant to be exhaustive or all-inclusive.

If your platform is NT or Windows 2000, install MSCS software on each node you intend to cluster. (You must leave some nodes unclustered if you deploy Central Dispatch for load-balancing or connection brokering.)

On the copy you made of Deployment Planning Worksheets fill out the section related to MSCS, so that you can refer to this during installation. You should have a pool of IP addresses to select from for the purpose of clustering.

Configure computers, disk arrays, and MSCS cluster software, making sure that cluster groups of disks have been created and made available.

Clusters should have at least two network/NICs. One network is normally dedicated to cluster heartbeat traffic, with the other used for general network access. The heartbeat network is not normally publicly accessible. The public network has the cluster IP addresses and carries cluster traffic, non-cluster traffic, and acts as a backup for the heartbeat traffic.

You should install and have running Siebel Gateway Name Server (without Central Dispatch Software).

You should have multiple running Siebel Servers within the same Siebel Enterprise. If you have multiple Siebel Servers running that are not clustered, these should be load balanced using Central Dispatch.

It is important to cluster the Siebel Database Server as it is a single point of failure. If you intend to cluster the Siebel Database Server, you should have installed it and it should be running within its own cluster node. The Siebel Database Server should be the first server to be clustered.

NOTE: You can cluster Siebel eBusiness Applications if the Siebel supported db platform has cluster support from the database vendor; see your database vendor documentation on how to cluster your database.