SSCC CPU Usage Policy

The SSCC has a large number of CPUs available through our various servers, but we also have over a thousand users. Roughly speaking, about 20 users share each CPU.

If a server is given a CPU-intensive job, and most statistical jobs are CPU-intensive, it will normally devote an entire CPU to the task. If a server has CPU-intensive jobs running on most or all of its CPUs, all users will notice significant performance degradation for even the simplest tasks. Windows servers can become completely non-responsive. The servers are a shared resource, and overloading them will affect everyone.

Thus we ask all our members to adhere to the following:

1. Only run one CPU-intensive job at a time on Winstat, or one job each on KITE and HAL. Jobs like text editors, email programs, etc. use very little processor power and you may feel free to run as many of these as you wish. Jobs like Stata, SAS, SPSS, other statistical programs, or user-written programs are a different matter.
2. Only run jobs on FALCON that require the very large amounts of memory only FALCON can provide.
3. Submit no more than ten jobs to Condor at a time, unless you anticipate that the jobs will be done in a very short period (on the order of thirty minutes or less) and there are enough CPUs available at the time you submit them.

SSCC staff regularly monitor the load on each of the servers. If we notice a user is running jobs in violation of this policy we will try to let the jobs finish if possible, while reminding the user of the policy. But if the jobs are causing problems for other users we may need to terminate them immediately.

This will not keep the servers from getting busy on occasion (remember how many users there are per CPU). But it does ensure that they will not be overloaded by a single user or very small group of users.