Simple end-to-end load and bottleneck monitoring for DB-based web sites
Posted
by T.J. Crowder
on Server Fault
See other posts from Server Fault
or by T.J. Crowder
Published on 2010-06-09T13:25:56Z
Indexed on
2010/06/09
13:32 UTC
Read the original article
Hit count: 469
What tools do you use / would you recommend for monitoring a Linux-based, DB-based website's servers for bottlenecks and load? The obvious goal being to know when growth has gotten to the point where it's necessary to scale up (or out) one or more of the bits and pieces because the current system won't be managing the load if an observed trend continues.
I'm looking for general recommendations based on standard Linux load metrics, disk I/O metrics, network I/O metrics, etc., but if specifics are helpful: It'll be Tomcat6 using APR (possibly with a Varnish or similar caching and balancing front-end), MySQL, and either Ubuntu 8.04 LTS or 10.04 LTS depending on timing.
I know about top
, vmstat
, iostat
, bwmon
and the like that collect and parse info from the /proc
file system (et. al.); and obviously MySQL provides a lot of queriable performance information. I could use those directly, probably automating periodic monitoring logs with scripts and such. But I have a suspicion that I'd be reinventing a wheel...
For example, Hyperic HQ seems to be along the lines of what I'm looking for. Others?
Meta: I tend to think of "recommendation" questions as needing to be CW because there's no one right answer, but I see a lot of these here that aren't CWs, so I haven't marked it as one. I'll happily do so if enough people think I should.
© Server Fault or respective owner