Optimum number of job servers - trial and error?

scheduling XI Scheduler
1

Hi,

I am trying to tweak our XIr2 configuration to see if we can improve performance and/or stability for our scheduled reports running at the weekend…we usually have 2000 or so (all deski).

We have 2 clustered servers running with the same BO configuration on each - we have 2 desktop intelligence job servers running on each, and each job server has “Maximum Jobs Allowed:” set to 20, so we can have 80 Deski jobs running at any one time.

Virtually all reports use the same data source and that datasource has a maximum of 40 concurrent active queries so the other 40 will just que up on the database and more will be “pending” in BO.

I want to reduce the “maximum jobs allowed” by half on each job server so the datasource will only ever get 40 jobs - the number of jobs it can actually run.

Does anyone know if I should be changing anything else? i.e. are the job servers connected to other servers that should be adjusted accordingly?
(Any other advice?)

Your help is appreciated.

Thanks

rob_bo :uk: (BOB member since 2005-08-04)

2

Only Job Server’s will have the “maximum jobs allowed” parameter, other servers do not.

I hope everything should work fine and don’t forget to stop / restart the services.

BO_Chief :us: (BOB member since 2004-06-06)

3

Some trial and error will still be required. A report may create more than one connection to the data source. Once instance of this possibility is if the report has sub-reports. I can’t remember about DeskI, but we ran into this problem with Crystal Reports.

For us though, this was only a small subset of our reports. We were able to assign them to only run on one job server which allowed us to still keep all of our job servers available for other reporting. This doesn’t sound like a feasible alternative for you though…