Filemaker stress test

[H.P.]

Does anyone know a good way to do stress test on the filemaker server? I need the server to work well with 25 users. I do not have access to 25 computers until deployment time. Do you know any good tool that will help with the stress test?

Also, I tried with 3 computers continuously running the script that will create new project + changing records. Server dies after 1 hour. Very strange.... does anyone else have the same problem?

[stanley]

HP:

Tell us a bit more about the server. What OS, drives, RAM, etc.

Also, a bit more about the script you ran.

-Stanley

[H.P.]

I found the reason why the server keeps dying every few hours. I am testing FM Server using the trial version. And the trial version will stop the server every 3 hours

Thank you for your help!

H.P.

[stanley]

HP:

If only every problem was so simple...

-Stanley

[Greg Hains]

Hi.

For the full/registered version, does anybody know of a good stress test for Filemaker 11 Server? Mine doesnt drop out every 3 hours, but it does hiccup long enough to get a connection timeout.

I have watched Task Manager on the server and it doesnt appear to be a load thing. One of my clients gets the same dropout so its related to the host (the server or the host's Internet connection). I can work on the host itself, but want to see how the FM server handles a large load.

Does anybody have a example database or routine that may do this?

Cheers,

Greg

[IdealData]

This is most likely the internet connection getting overloaded as 25 users is well within capacity of FMS.

Any test is subjective to the solution as all solutions are different. Graphic objects will create high demand on the internet connection as they have to downloaded to the client before being displayed. A way around this is to develop using the Separation Model where the user interface can be held locally and the hosted files are purely data.

[Wim Decorte]

I have watched Task Manager on the server and it doesnt appear to be a load thing. One of my clients gets the same dropout so its related to the host (the server or the host's Internet connection). I can work on the host itself, but want to see how the FM server handles a large load.

Can you describe "hiccup" and "dropout" in some more detail? What kind of messages are the users getting? What does the FMS event log say? Or the OS event logs?

As to stress test: it's really hard to come up with a generic one because a lot depends on the design of the system. If you have lots of filtered or sorted portals, sorted relationships, an overloaded graph, lots of non-FM graphic elements, etc then the same hardware will be able to serve less concurrent users than for a solution without those.

Go through your solution and isolate the slowest bits. Then get 5 users in a room and let them perform those slowest tasks at the same time over and over again. Enable client stats in the FMS console and OS performance monitoring.