On Mon, Nov 03, 2008 at 08:47:34PM +0100, Tobias Oetiker wrote: > except that normally multi argument updates are with 100s of > updates, so that should not matter much ...
>From the daemon to the RRD file, there are 100s of updates per file. However, when polling there is normally one update per file per "step" interval.. The poller --> daemon is the source of the small writes. If there is a poller that is coalescing 100s of updates before sending to rrdcached, it is essentially duplicating the functionality of rrdcached. >> [implementation of batch API removed] > this sounds fine, except that I wonder if it is necessary at all. I just wanted to open for discussion in case there was a lot of interest... I agree that it's not necessary via the API. The result may be cumbersome. > If someone wants to write 'for the daemon' they can easily talk to > to the daemon directly. In my current setup, I am connecting directly to the daemon. The protocol is simple enough to implement (thanks Florian), so implementing a client should not be a burden on anyone that needs extreme update rates. >From my initial list, that leaves only the small clean-up items. I should be done with those by the end of the week. -- kevin brintnall =~ /[EMAIL PROTECTED]/ _______________________________________________ rrd-developers mailing list [email protected] https://lists.oetiker.ch/cgi-bin/listinfo/rrd-developers
