Question: ideal graphite server configuration for high-volume environments?

[lukevenediger]

Hi,

My company's graphite server runs on an ubuntu 12.04 box with two cores and 4gb of RAM, with a SAN volume for storage.

Lately I've had to restart apache at least once a day because the dashboards get stuck waiting for graphs to render, but they never do. After a restart I'm able to load the dashboards quickly. At the same time, volumes on the system are growing and so this problem is only likely to get worse unless I scale it out somehow.

What has worked for you in your organisation? How did you scale your set up to handle increased query load?

Thanks!
Luke

[obfuscurity]

This really belongs on the mailing list or Launchpad Q&A. There is no one ideal configuration. The best thing you can do is to optimize for IO using SSDs or arrays of very fast disk. From there it just takes standard system engineering practices to identify the bottlenecks and adjust your configuration accordingly.

I have some examples up on obfuscurity/graphite-scripts#4 that may be helpful. We're also planning to document Graphite's clustering capabilities better, you can follow #816 for updates.