Anyone with a Cluster of WebFOCUS reporting servers (CLM)?

We have multiple environments, one with only one WF reporting server, others with three WF reporting servers. An HTML Composer page that has multiple dynamic controls takes almost twice as long to load when run in the cluster server environment. Is this possibly because the dynamic requests are run on the different servers and returned to the web server?

Just wondering.

This message has been edited. Last edited by: <Kathryn Henning>,

I have three chained list boxes dynamically populated by an external fex. Since I have to specify the fex in each of the three controls, may I assume the fex is called three times?

I have four other controls dynamically populated by external fexes.

I have a calendar control dynamically populated by an external fex, and also have specified an external fex to retrieve the min and max allowable dates.

9 fexes to populate the controls. Are these calls spread over the clustered servers? Is that a bad thing?

For this parameter screen I followed the rules and everything but the JavaScript was point-n-click.

Even with the cluster defined, you can aim a request at a particular component server in the cluster. So you can test your hypothesis, by (on the one hand) aiming them (the dynamic calls) all at the same server, and OTOH spreading them out over several.

If the underlying data lie on a relational DB server, perhaps responding to several requests coming on a single connection (as is the case in your Dev environment) works faster than responding to incoming requests on disparate connections.

Let us know (as you generally do).

I have a clustered environment with a CLM and I don't see any performance issues.

I do have a paticular portal with several controls as well as a Single Source Tree Control displaying product hierarchy information. It performs the same in our non-clustered environment as our clustered.

But if you do find anything, do please share. I am always looking at ways to improve performance.

Matt, thanks for the feedback.

Jack, are you suggesting I could hard code a IBIC_SERVER variable for each of the calls on the HTML Composer page, or set one for the whole page? I'll give that a try.

Hardcode, or Bang in values supplied in a covering fex

Bypassing CLM may give some performance benefit as you avoid "negotiating" with CLM which Reporting Server should run which .fex; in essence there are less gears to set into motion However fixing .fexes to run off of a particular IBIC_server defies one of the important purposes of CLM: High Availability.

The idea behind it is that if SERVER1 becomes unavailable due to emergency maintenance or a crash for instance, CLM would direct the requests to SERVER2 instead increasing the availability of your application while making it transparent to users.

Hopefully you can find some other alternatives to improve performance on that particular page without having to sacrifice HA but that balance is not always easy to achieve.

njsden: Of course. I only proposed it as a stratagem for testing the hypothesis, not for use in production mode. Didn't I make that clear?

Jack, you did make that clear. Thanks.