Recommendations for aggregating all metrics

[rodmur]

Hi,

Just I was just wanting to aggregate all values coming from collectd and then pass it along to a graphite setup. Just wondering if the following is a good idea or not:

aggregate 
  (\.*)
  every 300 seconds
  expire after 360 seconds
  compute avg write to
    \1
  send to portaldb
  stop
  ;

Or is the above a bad idea?

[grobian]

Not per-se, depends a bit on how many unique metrics names you're going to spool through this instance :)

It will need memory obviously, and it's going to take some cpu to do the lookups all the time. I would just try and see if it falls over or not.