Ad blocker detected: Our website is made possible by displaying online advertisements to our visitors. Please consider supporting us by disabling your ad blocker on our website.
Obviously keeping data for weeks/months/years is going to result in BIG files, but aside from the disk space being used are there really any appreciable downsides to keeping RRA data for extended periods of time?
Does is reach a point where it starts affecting the performance of Cacti?
Does it create any other problems that I may not have thought about?
With disk space so ridiculously cheap these days, I'm just curious if there's any good reason not to hang onto data for ages ... just because I can!
It's more of a performance issue with rrdtool, not cacti. Offhand, I don't know of the permanence costs with the increases storage size. The two obvious areas you'd be interested in would be rrdtool update and rrdtool graph commands.
If you want, you could do some testing with a normal and large rra files and report the differences. The rrdtool mailing lists would also be a good place to post this question.
I'm about to start looking into using cacti combined with rrdtool for this very reason. I made a custom page that graphed the usage of our student labs with rrdtool but now the scope of the project has changed and my manager wants to be able to examine the usage of a lab in a 24 hour window (5min polling) for any day in the past two years.
That's 210,240 queries to be stored and be accessible/graphable at any time...could be fun
You really should ask at the rrdtool-users mailing list. They are quite responsive, there.
I suppose, that with most versions of rrdtool, you may get some I/O overhead, but not a huge one, if any at all. Due to the nature of rrdtool files, rrdtool should fetch a single sector only (but OS file caching features may add more I/O. In this case, use latest rrdtool 1.3 that makes use of fadvise to avoid unnecessary read caching. Available for unix flavours only, AFAIK)
Reinhard
