[nycbug-talk] Cacti Sucks, So what do I replace it with.

[Pete Wright]

On Thu, May 05, 2011 at 12:18:56PM -0400, Mark Saad wrote:
> Talk
>   I have a good question for you. I started to hate cacti for a few
> reasons I dont want to get into.
> 
> I know that a few other trending / monitoring projects have reached
> critical mass have a good number of people using them.
> What do you recommend I move to.  Here are my requirements.
> 
> 1. SNMP Polling
> 2. RRD , SQLite, or Berklydb data storage
> 3. I don't want it to lower my tco or bake me a cake .
> 4. Flexible trend management.  (If I want to trend nfs read operations
> for 100 servers into one graph I should not have to jump threw hoops)
> 
> So people have pointed me to
> 
> 1. zabbix.com
> 2. munin-monitoring.org
> 3. ganglia.sourceforge.net
> 
> What are you using ?


i've found that ganglia can be pretty helpful for clusters of systems,
although it is agent based.


regarding your requirements i'd like to throw somethingout there that
violates #'s 1 and 2, although i think it will provide you with a system
that is more scalable and flexible moving forward.

http://graphite.wikidot.com/


I've used graphite in on very large clusters, and have used it to plot
all sorts of data and very high rates.  We found that while it does not
support snmp out of the box, it is trivail to get it setup to poll data
via snmp.  the flip-side is that since it does not rely on snmp you can
gather a wider array of metrics w/ less work required to get it versus
snmp.

for example, you could write a script that executes a SQL statement that
outputs interesting data from your RDBMS (how long has a given query
run, whats the status of my VACUUM metrics, etc..).  while i initially
found that this unstructured approach made my sys-admin senses tingle it
did provide us with some interesting opportunities i would not have been
able to easilly do purely via SNMP.

I've also found that the "whisper" fileformat that graphite data is
stored in is quite a bit more flixible, and more efficient, than rrd -
and most def faster and more scalable than sqlite or BDB.


-pete

-- 
Pete Wright
pete at nomadlogic.org