We are attempting to make a justification for adding more Data Collectors to PM. However, the sever CPU and memory don't reflect a need to add capacity. The DC admin page says we are at 80% and 81% DC capacity based on 500,000 items. So just need some clarification around what Broadcom recommends as a point at which we need to add capacity.
What does DC capacity in the GUI represent?
What does pause time represent in the Data Collector health page?
What does heap percent in the Data Collector health page?
Does QOS classmap collection effect Data Collector sizing?
Is QOS classmap collection taken into account in the sizing spreadsheet?
Release : 20.2
Component : IM Reporting / Admin / Configuration
The DC capacity is just based on element counts vs the set capacity. To see if a data collector needs either more resources or more data collectors added you should review the Data Collector Health page.
Check the heap % if the heap % goes over 80% then garbage collection will automatically occur as a rule of thumb if a data collector is running steadily over 70% then it has reached capacity for the memory on the DC and no more should be added. Another data collector or more memory maybe needed for peak times.
Another thing to look at is pause time this is the amount of time the app is paused due to garbage collection in java - aka reclaiming freed memory for use. If this is consistently above 10-20 seconds more memory or another data collector may be needed.
Note: To be safe double the amount of processors and memory on any Data Collector doing QoS classmap collection as the wizard only takes into consideration item counts and associated load, not the added overhead of QoS collection and processing.
link to the CAPM sizing tool: