Reserving servers for specific functions

You can configure servers in a cluster to perform specific functions such as batch processing and Exchange synchronization.By isolating processes to specific servers, you can help to improve system stability and minimize impact to online users in the event that there is a performance issue with a process.

You reserve a server by specifying one or more node types for the server node. Each node type represents a set of functions that a server can perform. There are three default node types available:

batch: Specifies that the server node can participate in server side processes such as object queue processing.
exchange: Specifies that the server node can participate in Exchange synchronization processing.
web: Specifies that the server node can participate in HTTP session load balancing.

For example, in a server cluster with two nodes, if you specify the batch and web types for one node, and the exchange type for the other, Exchange synchronization work will be isolated to the second node. If an issue then occurs in the Exchange node, it will not affect batch and web processes in the other node. By default, every server in a cluster is configured with all three node types, meaning that all servers handle processes equally.

You can also create your own node types and specify what processes are associated with each type. To do this, you edit the Core.ntypes file to create your custom node types. Processes are grouped together into component, channel, metaclass files, which you can edit to specify which node type is associated with each set of processes.

For example, in the default configuration, session manager and HTTP channel processes are both associated with the web node type. If you want split the processes so that they can be run on separate servers, you can create a node type for each process, then edit their component and channel files to specify the new node types.

You configure a server's node types on the System screen in NexJ Admin Console.