On Fri, Feb 20, 2009 at 12:36:41PM -0800, Ben Rockwood wrote: > Peter Tribble wrote: > > On Wed, Feb 4, 2009 at 8:22 AM, adrian cockcroft > > <adrian.cockcr...@gmail.com> wrote: > > > >> Don't write yet another performance stats collector / plotter, its been > >> done > >> to death. > >> > > > > It may have been done to death; has it been done properly? > > > > I've been playing with most of the candidates. One of the reasons for doing > > so > > is because I have in the past rolled my own and feel there must be a better > > way, > > and because the dreaded 'metrics' have cropped up again. > > > > What I'm finding is that each one does some things well. But none are > > anything like > > complete; all require considerable effort to make work anyway; and there's > > major > > investment required to fill in the gaps (or implement multiple solutions). > > > > Despite knowing that it's wrong, there's a distinct possibility that > > we roll our own again. > > > > I have to agree with Peter... there are many tools but I'm not certain > any of them are great. > > My two personal favorites are: > > * Zabbix: Client based and easy to extend. > * OpenNMS: Best I've found if you are an SNMP heavy shop.
I would add HypericHQ to this list. In our researches Hyperic made out better than Zabbix. > Too many monitoring solutions focus heavily on SNMP, which is why I've > been a fan of Zabbix. If we get a decent MIB that may change. For > instance, Zenoss is pretty, but its tilted heavily toward SNMP with > extending via a custom scripts being ugly and complex, imho. OpenNMS and Hyperic formed a partnership. Regards Przemyslaw Bak (przemol) -- http://przemol.blogspot.com/ ---------------------------------------------------------------------- Promocja w Speak Up. Angielski 50% gratis! Liczba miejsc ograniczona. Sprawdź!>> http://link.interia.pl/f205c _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
