When a node has the compute role, the Service Manager manages the containers on the node. A container is an allocation of memory and CPU resources. An application service uses the container to remotely perform computations on the node. For example, a Data Integration Service grid includes Node 1 with the service role and Node 2 with the compute role. The Data Integration Service process that runs on Node 1 runs a mapping within a container on Node 2.

A node requires the compute role when the Data Integration Service runs jobs on the node. When the Data Integration Service runs on a single node, the node must have both the service and compute roles. When the Data Integration Service runs on a grid, at least one of the nodes in the grid must have the compute role.

A node does not require the compute role when the Data Integration Service does not run jobs on the node. In this case, you can disable the compute role on the node. However, because the container management function of the Service Manager is a lightweight process, enabling or disabling the compute role does not have a performance impact.

When you disable the compute role on a node, you must specify whether to stop, complete, or abort computations that might be running on the node.