About memory footprint of agent process
- Last Updated: February 11, 2026
- 2 minute read
- OpenEdge
- Version 13.0
- Documentation
This topic describes what contributes to the memory of the PAS for OpenEdge agent process
and how the operating system manages memory. Later, the operating system drills down
into why the memory reports generated using operating system utilities like
top (in Unix) show no drastic change in memory even after trimming
the ABL sessions.
- To correct the size of machine resources parameters like CPU, memory, swap, and others, to ensure that the server meets the response-time requirements without over-spending on resources.
- To monitor and control the size of the PAS for OpenEdge agent process and
prevent it from getting killed by
Out of Memory Reaper, which may be running on the machine. - To control the size of the PAS for OpenEdge agent process from impacting: the memory of the machine and the performance of other critical business applications running on the same machine.
PAS for OpenEdge agent tracks its calls to the malloc API
(malloc, free, calloc,
realloc, strdup) for self-reporting its known
internal memory consumption. However, the reported amount of memory used often does not
match what the OS tools report for the PAS for OpenEdge Agent process.
One of the main reasons for the difference is that the
malloc API adopts an optimistic memory allocation model to strike a
balance between maximizing speed and minimizing memory fragmentation when presenting
virtual memory to the agent process. The optimistic memory allocation model prioritizes
memory allocation and performance over the reduction of process memory size.
- ABL sessions and overhead memory in the PAS for OpenEdge agent. For more information, see About factors accounting for total memory in OpenEdge.
- Shared memory and swap. For more information, see About factors in the operating system that account for memory usage.