[plum] Plum under load?

[Aaron Longnion - hotmail]

Hi Adam and David -   I'm very interested in any information you can provide about the performance implications of using the Plum framework under very heavy load.  At my new job, I may want to start convincing folks to use Plum in certain instances, especially new development.  I know they're going to want to know how it performs under intense load (perhaps eBay-type load within a few years, if everything goes well).  We'll really want to scale our applications (application, web, and database clustering will most likely be involved).  We would use some pretty powerful app and DB servers, with a mix of Win 2K and Win 2003.  The DB will be SQL Server 2000.  So...   1.) What kind of load testing have you done, with what tools, with what hardware/software specs, and what were the results?   2.) How do you suggest setting up Plum in a CF MX 6.1 clustered environment (possibly multiple machines, and probably multiple instances of CF)?   3.) Normally, with a small cluster (2 or 3 app servers), we would use Client variables with DB storage.  So, would you do this in a larger cluster under very high load?  Perhaps we could use a seperate DB machine just for Client session management?   4.) Are there performance implications with the way Plum puts so many CFC's in the Application scope?   5.) I assume you would recommend switching as many of the <CFQUERY> tags to Stored procedure calls?  By the way, one recommendation would be for an easier way to just make Plum use all sprocs instead of inline <CFQUERY> code in the V1.1, as if you guys don't have enough to do...  :)   6.) Posting/documenting the results of performance metrics for Plum under common configs with intense load would be very helpful if anyone has the time and setup to do it.   7.) We might also move to CFMX 7 soon, so any information on how Plum performs under load with 7 would be helpful.   Thanks much! Aaron