The Agent Activity tab on the Database Manager Summary drilldown provides a summary of agent activity on each database and partition in the system.
The Agent Activity tab contains the following graphs, which keep track of agent activity over consecutive monitoring intervals:
High rates depicted on the graphs for the following activities might be indications of performance degradation and inefficient use of system resources.
In the Active Agents graph, comparing the number of agents from the agent pool to the number of agents created from the empty pool can help you determine if the agent pool is being used efficiently.
A high number of dynamically created agents from the empty pool can indicate two things:
The NUM_POOLAGENTS parameter should be increased.
The overall workload for this partition is too high and should be corrected by lowering the number of coordinating agents or reducing the workload.
A low number of dynamically created agents from the empty pool can indicate that NUM_POOLAGENTS is set too high. What this means is there are seldom used agents residing in the agent pool that are unnecessarily using system resources.
Both the local connections and remote connections statistics can help you gauge a need to adjust MAX_COORDAGENTS. If there are more connections than MAX_COORDAGENTS then applications that do not have a coordinating agent are inactive until there is a coordinating agent to service that connection. Because of this, the MAX_COORDAGENTS can be used to fine tune the load on the database server.
The idle agents graph can be used to fine tune the value of the NUM_POOLAGENTS parameter. Having idle agents available to service requests for agents can increase performance of the database server. In environments where there are many applications connected, it is generally recommended to increase the NUM_POOLAGENTS value to avoid the unnecessary overhead of creating and destroying agents.