Hi Benjamin, On Wed, Mar 06, 2013 at 07:01:29AM +0000, Benjamin Wang (gendwang) wrote: > Is there any potential bottleneck for this scenario? Do you have any > experienced data about the collectd performance?
your bottleneck is going to be the I/O performance when writing RRD
files to disk. With decent hardware (6x 15k disks in a RAID-10, e.g.)
you can expect between 100 and 200 updates[0] per second. The "rrdtool"
plugin caches and batches these writes to help matters, or you can use
the external "rrdcached" which decouples config changes and cache
flushes. I suggest you read through the "Inside the RRDtool plugin"
article[1] which describes the problem in detail and tells you which
mechanisms were implemented to cope with them.
In my experience, the network plugin itself is not a bottle neck.
Best regards,
—octo
[0] In this context, one "update" means one read/update/write cycle that
writes one or more values to a single RRD file.
[1] <https://collectd.org/wiki/index.php/Inside_the_RRDtool_plugin>
--
collectd – The system statistics collection daemon
Website: http://collectd.org
Google+: http://collectd.org/+
GitHub: https://github.com/collectd
Twitter: http://twitter.com/collectd
signature.asc
Description: Digital signature
_______________________________________________ collectd mailing list [email protected] http://mailman.verplant.org/listinfo/collectd
