I am migrating to a new LDAP product and am concern about the load on the new system. Both my old and new environment are behind the same F5 BigIP systems. I have done my best at replicating this load, but I was wondering if anyone had any good ideas to use production traffic and maybe a clone pool (or fancy irules) to run a real load scenario. I’m not sure if this would work or not. Any advice would be greatly appreciated!
There aren’t (IMO) any really truely good LDAP load generators out there…
I did manage to find a small C program last year that was (A) configurable, and (B) Fast for some testing work I was doing with a client last year… It was written in C (Perl was too slow), and needed a couple of alterations to work with Solaris ‘not quite Unix’ 10 instead of Linux, but ran OK after that.
The software was called (IIRC) charge… If you have problems finding it, drop me a line and i’ll see if I can dig out the copy I was running and upload it.
Oh. If you have plenty of CPU, then you could write a small perl script to read the activity logs from your current servers and replicate the traffic to your new ldap infrastructure… Lack of CPU was the reason I abandoned that.
I wrote my own generator with Perl and was pretty successful, but my old ldap infrastructure doesn’t have very good logging/auditing capabilities, so I wasn’t able to get a 100% true representation of my traffic - only a basic idea. Thats why I wanted to see if I could use clone pools to toss real traffic at it without consequences.
Yes, I believe you could use a clone pool (Just make the new pool the clone in your existing VS). (Although they’re really targeted at IDS systems)
Also, make sure that you’re just copying client-side traffic only (Your new servers aren’t interested in server side stuff from the old pool), and away you go.