When concurrent users have too many query server processes, memory contention (paging), I/O contention, or excessive context switching can occur. This contention can reduce system throughput to a level lower than if parallel execution were not used. Increase the PARALLEL_MAX_SERVERS
value only if the system has sufficient memory and I/O bandwidth for the resulting load.
You can use performance monitoring tools of the operating system to determine how much memory, swap space and I/O bandwidth are free. Look at the run queue lengths for both your CPUs and disks, and the service time for I/O operations on the system. Verify that the system has sufficient swap space to add more processes. Limiting the total number of query server processes might restrict the number of concurrent users who can execute parallel operations, but system throughput tends to remain stable.