Hi Liuz,
In the example directory of the enterprise manager, there are two management modules: MOM_Infra_Monitoring_MM.jar and collector.jar. Create a copy of the collector.jar for each of your collectors. Deploy the management modules to your MOM. Customize the agent management module expression for each module. Add your collector management modules alerts and dashboard links to the MOM's dashboards. Open the dashboard "MOM Infrastructure Overview"
Now, with everything set up, you can start to understand what metrics are in the APM to help address your issue.
Ping could be three factors, network, destination system, source system. Now, with the APM ping, as Hiko mentioned it isn't your typical ping. The MOM will connect to the collector's message queues, post a message and wait for a response, so if the MOM's queues/processes are too busy to handle the response or if the collector's queues/process are too busy then you will get an increased ping time.
Now with the MOM Infrastructure Overview, there there a branch that has "Performance", it is a good place to start.
On the Collector's - Collector Performance dashboard, there is harvest duration, smartstor duration, number of event inserts, query performance and CPU & Memory.
Try to correlate the ping (Dashboard - MOM Cluster Ping Time) to the enterprise managers and see if there is anything that might help determine the root cause of the problem.
Hope this helps,
Billy