← Back to team overview

graphite-dev team mailing list archive

Re: [Question #217012]: Metric aggregation for sub second frequency

 

Question #217012 on Graphite changed:
https://answers.launchpad.net/graphite/+question/217012

Description changed to:
Thanks for making Graphite!

I want to collect 6 metrics for 20.000 individual applications, coming
in from about 100 servers with a varied frequency (1-60/sec) and I
wonder if I could use your elegant software to do it. I noticed that
multiple metric submissions within the smallest bucket time get
overwritten.

In 2010, Matthew proposed [1] to put a custom made bucket/aggregator in
front of carbon. I wonder if anyone has already written a generic one in
the meantime ;-) In my situation, I would like to sum everything in 1
second buckets before feeding it to carbon. I don't think I would be
able to use carbon-aggregator.py for that, but please correct me if I'm
wrong.

Background info: I collect CPU & I/O stats per page request. Since they
vary a lot, I wouldn't trust only a fraction of the samples  so I want
to sum everything. At the moment I do this with a custom made system
based on RRD, but it's not very flexible with querying data. After
playing around with Graphite I like it a lot, so hopefully we'll be able
to migrate.

Thanks!

[1] https://answers.launchpad.net/graphite/+question/119934

-- 
You received this question notification because you are a member of
graphite-dev, which is an answer contact for Graphite.