What does a memory leak look like on a graph? How did you detect it?

Memory leak on a graph usually manifests as a continuous increase in the application's memory consumption without subsequent decrease, even when the load stabilizes or decreases. It may look like an upward curve that does not return to the baseline.

To detect a memory leak:

1. Memory monitoring: Use profiling tools (either built-in to the ERP system or external) that create graphs of memory usage over time.
2. Comparison with expected behavior: If memory does not release after load peaks, it indicates a leak.
3. Memory dump analysis: Take memory snapshots (heap dumps) and analyze objects that are not released.

Example: During testing of an ERP system, it was observed that after a series of operations, memory grows and does not decrease, indicating retention of references to objects that should have been deleted.

Thus, a memory leak on a graph is a constant growth of used memory without reduction, and detection requires monitoring and analyzing the application's behavior over time.