On Thu, 24 Aug 2006, Jim Trocki wrote:
> this probably requires much more thought and a better understanding of the
> usage scenario.
so here's the info that would be useful:
-how many services on each of the 1000 mon servers
-what their intervals are
-how many failures per sec would you typically see
then you can figure out what kind of processing load the main mon server which
receives the traps needs to handle, in terms of "no problem" traps per second
and "failure" traps per second.
long ago i used to do performance tests by running mon on a sparcclassic with
linux, to see how it held up with crappy processing power, and for the tests i
did it was able to handle like a thousand "all is ok" monitors every, oh i
don't recall, 10s or so.
the real trick is to get a good estimate of the workload you will actually
be doing, with all the details of the alerts (email? pager? whatever) that
you expect to be processing.
_______________________________________________
mon mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/mon
