Rrdcached has more effect as you increase the caching time. IE, if you set it to cache for 30min, and your update for a given RRD file is every 5min, then you should end up making about 1/6th of the disk IO (actually a bit more than that due to RRA rollups, but its a good rule of thumb)
The people with huge update rates seem to be doing this -- 1. big cache times, like an hour or two 2. caching in memory (and a lot of memory available to use) 3. journal (which you need if having a big cache time) on very fast disk 4. high update-to-rrd rate, IE updating one rrd every minute is more efficient than updating 5 rrds every 5min if using rrdcached This is how we run rrdcached (~60 updates per second) /u01/rrdtool-1.4/bin/rrdcached -b /u01/rrddata -s mrtg -m 0666 -z 60 -p /var/tmp/rrdcached.pid -t 4 -j /u01/rrddata/journal -w 1800 -O -a 6 -l unix:/var/tmp/rrdcached.sock -l rrdserver.auckland.ac.nz -l localhost We only cache for 30min ( -w 1800 ). Note we increase the write threads ( -t 4 ) from the default of 1 which might be a good idea for you to consider -- you may be bottlenecking on this? Steve Steve Shipway ITS Unix Services Design Lead University of Auckland, New Zealand Floor 1, 58 Symonds Street, Auckland Phone: +64 (0)9 3737599 ext 86487 DDI: +64 (0)9 924 6487 Mobile: +64 (0)21 753 189 Email: s.ship...@auckland.ac.nz Please consider the environment before printing this e-mail
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ rrd-users mailing list rrd-users@lists.oetiker.ch https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users