Because the distributed
in-process scheduling server and the JMS queue manager are automatically
configured, the default configuration values are acceptable for most
situations. However, you can use these tips to optimize the performance
of distributed in-process scheduling:
-
If flows are not completing because
of competition with new flows, reduce the value of the
Max running flows field on the
Operation
Options dialog box.
-
If the distributed in-process scheduling
server or the job runners cannot keep up with the number of jobs running,
reduce the value of the
Max running jobs field
on the
Options tab.
-
If too much disk space is being
taken up by flow histories, reduce the value of the
Max
histories per flow field on the
Options tab. You also might want to reduce this value if users do not want
to see a large number of histories.
-
If your flows need to be highly
responsive to file changes that are used as file events, reduce the
value of the
File trigger/condition check frequency field on the
Operation Options dialog box. If
your flows contain a large number of file events, you might want to
increase the value in order to reduce the amount of time that the
system takes to poll the file system for changes.
Use these tips to optimize
processing of jobs that use the default job runner (jobs that are
executed by a SAS process):
-
If the jobs that are scheduled
are CPU-intensive, lower the value of the
Max simultaneous
jobs field on the
Options tab
of the
Properties dialog box.
-
If you have defined multiple distributed
in-process scheduling servers, decrease the value of the
Max simultaneous jobs field on all server definitions.
-
If you want the job runner to check
more often for jobs that are waiting to be run, reduce the value of
the
Polling frequency field on the
Renderer Options dialog box. Increase this value if
you want to reduce the amount of overhead used by the job runner.
-
If you want to increase the amount
of time before getting a new job, increase the value of the
Max wait time field on the
Renderer options dialog box.
-
If the jobs are not CPU-intensive,
the job runner can start processing new jobs more quickly, so you
can increase the value of the
Job count before wait field on the
Renderer options dialog box.
-
If the jobs are CPU-intensive,
increase the wait time for each job by increasing the value of the
Base wait time field on the
Renderer options dialog box.