Workload management allows system administrators to monitor
and adjust the speed that messages are processed, as well as controlling
the actions taken on unresponsive flows and threads.
Before you can use workload management functions, you must ensure that the publication
of events has been enabled and that a pub/sub broker has been configured. For more
information, see Configuring the publication of event messages and Configuring the built-in MQTT pub/sub broker.
The following topics explain the various options available under workload management:
- Message flow notification
A common requirement is to be able to monitor the speed at
which IBM® Integration Bus processes messages. Workload management allows the system administrator
to express a notification
threshold for individual message flows deployed. An out of range
notification message is produced if the notification
threshold is exceeded. A back in range
notification message is produced if the notification
threshold later drops back into range.
For more information, see Configure a message flow to cause notification threshold messages to be published.
- Setting the maximum
rate for a message flow
The system administrator can set the maximum
rate that an individual message flow can run at. The maximum
rate is specified as the total
number of input messages processed every second. When set, the number
of input messages that are processed across the flow is measured.
This measure is irrespective of the number of additional instances
in use, the number of input nodes in the message flow, or the number of errors that
occur. If necessary, a processing delay is introduced to keep the
input message processing rate under the maximum flow rate setting.
For more information, see Setting the maximum rate for a message flow.
- Setting additional threads to service a message flow
- You can set the number of additional threads that the integration node can use to service a
message flow, and specify whether the additional threads start when the flow starts. For more
information, see Workload management policy.
- Unresponsive message flows
Allows you to specify and monitor the maximum amount of time
that any message flow is allowed to
process a message for, and to specify an action to be taken if the
timeout is exceeded. Additionally, manual requests can be made to
stop a message flow by restarting the
integration server. For more information, see Unresponsive message flows.
- Workload management policy configuration
-
Instead of defining properties on the message flow, you can specify a workload management policy within the Integration
Registry for a message flow, so that message flows can refer to the policy to
find properties at run time. You can change the values of attributes
for a policy on the integration node, which then affects the behavior
of a message flow without the need
for redeployment. The workload management policy
encompasses all of the properties available under workload management in one place, including the notification
threshold and maximum
rate, and therefore allows for
easier tuning of message flow performance.
For
more information, see Workload management policy.