Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I am attempting to do a bit of a post-mortem on a recent launch and could use some expert input.

A site that I've had running for a bit had an anticipated spike in traffic due to an announcement. The site is an ExpressionEngine (php/MySQL) site running on a MediaTemple DV 4GB server.

Prior to the announcement, I hit the server with Apace Benchmark. I don't remember the exact settings I used, but it was something like 1000 requests with a concurrency of 30. When I did this I noted that the total processor usage in Top hovered around 60-70%. Also, while AB was doing its thing I attempted to pull up the site and it was fine.

Then the announcement happened. The server got totally crushed and basically didn't load the site properly for about 24 hours. During this time Top said the processor usage was around 20%.

What I can't figure out is how it was fine during my AB test but not during the real traffic spike. Granted, my AB test wasn't loading static assets, but the processor usage was way higher than it was when the traffic spiked, and it was fine.

Any ideas about this discrepancy? Thanks so much for any wisdom.

share|improve this question
    
Are you sure the bottleneck was the server and not, say, the internet pipe? –  Shane Madden Feb 14 '12 at 4:30
1  
Any chance you have sar installed on the machine, and can have a look at the network traffic coming in during that time? –  Decado Feb 14 '12 at 5:51
    
Unfortunately, no, I don't have sar setup. Sounds like a good tool for next time though. –  Matt Fordham Feb 14 '12 at 16:40
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.