I am not sure if anybody else has encountered this problem, still any suggestion on how to mitigate /troubleshoot the problem will be very helpful:
We have TM1 running on Windows server and there is an external application pushing data into the main cube using relative proportional spreading. The spreading is performed through REST API call and it's a batch request that runs 12 threads at a time using one time login.
The spreading performance is spectacular when the TM1 instance is restarted, but after a number of spreads the performance starts degrading and at some point it becomes unviable and restart of TM1 instance is the only option. Although the RAM utilization goes up after each spread, we can see that server has more than 100 GB RAM available, so I assume server memory is not an issue. In terms of numbers initially it takes around 1 min for 3000 * 12 threads , but later on it takes around 20 minutes.
Any suggestions as to how to troubleshoot and find the root cause would be quite helpful.
Thanks