Answer
An Admin Server heap error alone usually isn't what causes a full application outage — that's more typically tied to the Managed Server running out of heap, so check the Managed Server and HTTP server logs as well, not just the Admin Server's. Before assuming more memory is the fix, take heap dumps from the environment while the issue is reproducible and analyze them for a memory leak — if there is a leak, simply increasing the heap allocation will only delay the same crash, not prevent it.