[nycbug-talk] Cacti Alternatives?

Lance Laursen lance at demonware.net
Thu May 13 20:25:16 EDT 2010


On Thu, May 13, 2010 at 5:11 PM, Matt Juszczak <matt at atopia.net> wrote:

> Hi all,
>
> I'd like to set something up like cacti, but I'd really like to automate
> the creation of the graphs and templates with either puppet (which talks to
> LDAP for nodes) or LDAP directly.  Just like powerDNS is a great alternative
> for bind that allows an LDAP backend, I'm looking for an alternative to
> cacti (or a way to make cacti do what I want) that would allow me to use
> puppet, or LDAP directly, to add graphs to cacti and manage them.
>
> Does anyone have any ideas?  What have others done?  It seems like one
> could always write a process that would read data from LDAP and execute the
> command line scripts for cacti (or have puppet execute the command line
> scripts for cacti based on node data in LDAP).
>
> Thanks!
>
> -Matt
>


Since Cacti does not scale, and stores all of its template/device/whatever
information in a database, I would take a serious look at Graphite (
http://graphite.wikidot.com/). "generating" new graphs is as simple as
modifying a URL, and it scales to hundreds of servers per instance
(thousands if your polling and storage is efficient, depending on how many
people will be looking at graphs at once), something that is an awful thing
to manage in Cacti. Unfortunately you do not get the polling engine of
Cacti, but if you are generating graphing information by other means then
that won't be a problem. If you do indeed need the polling engine, I'm
afraid I don't know of any good alternatives.

Unfortunately the "run-once" nature of creating graph URLs (or generating
templates & adding to graph trees in cacti) falls outside of puppet's
purpose, but i'm confident any work done against graphite will be cleaner
than using cacti's cli tools.


-- 
Lance Laursen
Demonware Systems Engineer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nycbug.org/pipermail/talk/attachments/20100513/9b6f5585/attachment.html>


More information about the talk mailing list