Report Service Tuning

[Dev1975]

Hi all,

This may seem like a strange/silly question but I have to ask it anyway...

I've got Cognos 10.2 installed on separate servers (CM, Dispatcher and GW) all deployed to WAS. The Dispatcher server has 12 CPU processor cores and the setting for "Maximum number of processes for the report service during peak period" is set to 7. According to IBM's tuning doc, this should be set to 2 x CPU Cores, so this should be set to 24.

Would this current reduced number of processes solely contribute to Dispatcher timeout errors, or would other factors, such as WAS tuning, come into play as well? The Dispatcher server has 20gb of RAM installed and also runs PowerCube builds during the early hours. The Dispatcher timeouts usually occur during the course of the day when the environment is used to access reports.

Appreciate any info/input on this.

Cheers!

[MFGF]

Hi all,

This may seem like a strange/silly question but I have to ask it anyway...

I've got Cognos 10.2 installed on separate servers (CM, Dispatcher and GW) all deployed to WAS. The Dispatcher server has 12 CPU processor cores and the setting for "Maximum number of processes for the report service during peak period" is set to 7. According to IBM's tuning doc, this should be set to 2 x CPU Cores, so this should be set to 24.

Would this current reduced number of processes solely contribute to Dispatcher timeout errors, or would other factors, such as WAS tuning, come into play as well? The Dispatcher server has 20gb of RAM installed and also runs PowerCube builds during the early hours. The Dispatcher timeouts usually occur during the course of the day when the environment is used to access reports.

Appreciate any info/input on this.

Cheers!

Hi Dev,

It's highly likely this is the root cause of the timeouts if you have high demand for running interactive reports. The maximum number of concurrent interactive report requests the server can handle is a multiple of the number of processes and the number of connections. If you have 7 processes, each having 1 high affinity and 4 low affinity connections, your instance can potentially run 35 interactive reports concurrently. If there are more requests than this coming in, they will start to get queued. The queues have a defined timeout, after which the users will see a timeout message displayed, and it sounds like this is your situation?

If you increase the number of processes from 7 to 24, your server could then potentially run 120 interactive reports concurrently - almost four times the current limit - before any queuing begins.

Cheers!

MF.

[Dev1975]

Hi MFGF,

Thanks for your response, this was pretty much what I suspected was happening.

Cheers,
Dev