Performance tuning best practise

[Cinu8]

hi guys
any body have any kind of meterial or can any body give the
best practice steps for performance tuning for cognos8 BI

thank you in advance

cinu

[ducthcogtechie]

Rule of thumb is 90% of the performance is in the model and 10% in the server settings. The developer is king when it comes to that.
The benchmark for a report is 1 user on the server running a report. Just add a few seconds to that number, just in case a new bibus is spawned for a new report.
Adding cpu's and memory will not make that report any faster. It will just scale the server more for more users.

When reporting off powercubes, have it located on the same server as the BI dispatcher resides. If that means multiple copies, so be it. You get performance and redundancy. Do not put hese on \\server\cubes or something like that.

A default C8 install assumes a dual cpu server (2 bibus processes with 2 low affinities each) and 4 gig memory, thus a max of 8 concurrent reports.
Only change these values up when you have quadcores (or 4 duals), or extremely small and simple reports.
If you run it on vmware with a single core and 2 gig memory, make sure to monitor closely, as even the default settings may be to agressive, and need to be scaled down.