Re: [Observium] is_file() usage for rrd file existance

I have some proof-of-concept code that talks 'rrdcached' protocol for CREATE/UPDATE/FETCH and persists the data into OpenTSDB which is ontop of a hbase cluster, so can scale pretty much horizontally (ie more nodes=more metrics).

Well, that sounds interesting.

How does it work, exactly?

Does rrdtool get/give the data to the "cache", and never look at the disk?

How does the single-node performance compare to rrd?

adam.