Are there any mods/tweaks available to help prosper function quickly with a high traffic site? I drive about 150k clicks per day, spread out between 10,000-20,000 keywords, so the analyze/keywords.php feature seems to run very slowly with just 2 days of data already. I have a feeling running reports containing a week or a month would crash the server, and those long term reports are the most important to me. Any ideas?
My server specs:
Size: 'AMD Opteron(tm) Processor 6... Image: CentOS 5.5 64bit with Cpane...
CPU: Quad Opteron x8-6128 Speed: 2000MHz RAM: 12285MB
CPUs: 4 Physical CPUs Cores: 32 Total Cores RAID: Level 10
Disks: 4 Size: 545GB Type: SAS
Running php 5.3.8 and mysql 5.1.56 with memcache installed.
Help get this topic noticed by sharing it on Twitter, Facebook, or email.
we have a few tweaks (partitioning) on the database that will speed things up. Will probably roll that out on the next minor upgrade.
The other option (which may not be ideal) would be to archive your data in another database and clear out your clicks every few days.
Interesting, I am facing the same issue, I just posted about it in another post, what are we supposed to do about this. Is prosper 202 simply not for high volume users? archiving the data is defiently un-desirable as how are we to run long-term reports or what about conversions that take days to happen after the click.
I would like to ask a favor of you, please get the stats for what your cpu and other hardware load is while you are running reports.
I believe you will see that you are not even taxing your server one bit and its an issue with how prosper 202 implements mysql. See my post for more details. You can try playing with the partition's yourself, you don't really have to wait for any update from the dev's. I have the entirety of the technical explanation on there
you might even be able to squeze some more performance out of it if you go beyond what I did and set the partition period to something like 6 hours but this would limit how much historical data you could hold.