In the book "Monitoring With Ganglia" there is a discussion of scaling Ganglia
to large installations which focuses almost entirely on scaling I/O. I've
converted our installation from using gmond to using sflow as documented in
chapter 8 so as to reduce the effect of Ganglia monitoring on our network, and
have switched to using rrdcached, but now I'd like to lower our I/O by
adjusting sampling times and I find no recommendations for this in the book.
I assume that I'll need to completely redefine the RRDs if I lower the sample
rate, but I'm not clear on what I need to do exactly between hsflowd (running
in the hosts), gmond, gmetad and possibly gweb. (I'm guessing I need to match
the polling rate in hsflowd with the polling rate in gmetad). Can someone offer
some advice for scaling Ganglia by adjusting sample rates for large
installations (~1000 nodes)?
------------------------------------------------------------------------------
Symantec Endpoint Protection 12 positioned as A LEADER in The Forrester
Wave(TM): Endpoint Security, Q1 2013 and "remains a good choice" in the
endpoint security space. For insight on selecting the right partner to
tackle endpoint security challenges, access the full report.
http://p.sf.net/sfu/symantec-dev2dev
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
