Ideal sequence of jobs in a sequence

splayer · Post by **splayer** » Wed Feb 04, 2009 1:10 am

I have 30+ jobs and none of them are dependent on another. Some of them take a long time to finish while others finish in seconds. What would be the ideal way to place them in a job sequence so that performance for the overall sequence is best?

mk_ds09 · Post by **mk_ds09** » Wed Feb 04, 2009 2:50 am

L ->
| M
L -> | Max
| M
L -> | Max
| M
L - >

L :- Job takes minimum time...
M :- Medium time
Max :- Max time..

If you can have a design to start the jobs which takes less time simultanously and the job which takes the long time can run when there is less processing to do !

It will help in such way that the job which will require maximum time will have all the resources to use !

ray.wurlod · Post by **ray.wurlod** » Wed Feb 04, 2009 3:31 am

What do you mean by "performance" in this scenario?

splayer · Post by **splayer** » Wed Feb 04, 2009 9:37 pm

By performance, I mean I want the entire sequence to finish as fast possible.

kduke · Post by **kduke** » Wed Feb 04, 2009 10:08 pm

I would not put 30 jobs in one sequence. I would run in multiple sequences. Most of this is trial and error. You need to figure out how many jobs can run at the same time. The idea is keep the server at capacity without paging. Monitor your server.

Usually some jobs have to run before others. Like a parent table needs to load before a child table. Figure that order out first.

DSguru2B · Post by **DSguru2B** » Wed Feb 04, 2009 11:02 pm

Considering these are parallel jobs I would take Kim's advice on monitoring the server load. Do a bunch of runs with different sequence. Hey if your server can take it you can create multiple sequences and fire all of them at once if there is no dependency.

ray.wurlod · Post by **ray.wurlod** » Thu Feb 05, 2009 1:11 am

Since you have version 8, you can use the resource estimator to indicate the likely resources to be consumed by each job. Then assemble them so that any dependencies are addressed, and the total demand for resources remains under 100% of available resources.