On Wed, 23 Jan 2008, Matthias Blankenhaus wrote:

> 
> 
> On Wed, 23 Jan 2008, Vikrama H T wrote:
> 
> > 
> > Hi ,
> > 
> > I have gone through lots of mail chains in Ganglia-general forum regarding 
> > the performance of Ganglia with Scalability up to 1000+ nodes.
> > 
> > I am also planning to test Ganglia scalability up to 1000+ nodes .
> > 
> > I just wanted to know is any body have already tested this and documented 
> > performance metrics ?
> 
> I ran Ganglia 3.0.x on 2000+ nodes without seeing any issue.  I used a 
> hierarchical approach one grid spanning several cluster.  Each cluster 
> contained roughly the same amount of nodes.
> 
> > 
> > If so please give me those performance metrics.
> 
> The gmond was running under 1% CPU resource utilization using roughly 5MB 
> res. memory.  Gmetad was using around 8% CPU every 1 secs and roughly

Typo: I meant to say every 15 secs.

> drawing 13MB of res. memory. 
> I did not attempt any performance tuning on Ganglia.  I am very sure 
> that these numbers can be reduced even more.
> 
> > 
> > Also what are issues you faced during this 1000+ node testing ?
> 
> I did not experience any issue at all.  However, I did perform these
> optimizations, just in case: I did use the unicast model to reduce the 
> network traffic.  I also moved my RRDs to shared memory to reduce the
> load on the head node.
> 
> > 
> > What was the Polling interval configured at Gmetad for polling each Gmond?
> 
> Default: 15 sec.
> 
> > 
> > What would happen if we have configured 1 sec Polling interval for polling 
> > each Gmond for 1000+ Nodes.?
> > 
> > What could be the Latency of Metric collection from Gmetad with 1 sec 
> > Polling interval for polling each Gmond for 1000+ Nodes.?
> > 
> > Please help if any of you have answers for these queries.
> > 
> > Thanks and regards,
> > Vikrama
> > 
> > 
> > 
> > 
> > -------------------------------------------------------------------------
> > This SF.net email is sponsored by: Microsoft
> > Defy all challenges. Microsoft(R) Visual Studio 2008.
> > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> > _______________________________________________
> > Ganglia-general mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/ganglia-general
> > 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Ganglia-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/ganglia-general
> 

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general

Reply via email to