I'd recommend double-checking how your apache bench tests differ from your real-world user behavior. It sounds like there are some differences in the two and it's a common thing I've seen when folks do load testing before a big launch. Check that it exercises all the same things and is not just loading a static page or something that doesn't get into the database or other systems. Also if the problem is disproportionately impacting users on your wireless & WAN networks, try running the tests from your wireless & WAN networks.
-T On Nov 14, 2014, at 4:50 AM, Jeffrey J Ramsay <[email protected]> wrote: > I used the Apache Benchmarking utility to test our performance and we're able > to push through 5k requests with a concurrency of fifty and there was no > problem, but the system stops servicing requests after there are several > hundred WAIT requests under real world conditions. We've found many of the > latent WAITS are for Wireless and WAN connections. -- To unsubscribe send an email with subject unsubscribe to [email protected]. Please contact [email protected] for questions.
