www.jonsDocs.org.uk


Add your comment

This blog isn't allowing comments at this time.

Previous Comments


Cacti

On Wed 22nd February 2012, 16:09 Jonathansays:

Apologies, I dropped the ball somewhat but here is the promised post about Cacti.

In my previous post I discussed Smokeping and how it can help monitor your network / DNS resolution times / Internet speed etc. Generally, one tool doesn't do everything and often they're complimented by others and Smokeping was no exception, it's complimented brilliantly by Cacti (and vice versa).

Cacti uses the SNMP protocol to monitor network devices - switches, printers, servers - and then produces a funky graph that shows how different variables change over time.

At a number of customer sites I and a colleague have installed Cacti to determine bandwidth usage. It's not uncommon for a company network to have a 1gbps backbone but before deciding to pay out thousands for bigger backbone it would be beneficial to see what the usage is. If nothing else, this may justify (or obliterate) the IT department's business case for the cash.

When monitoring switches, Cacti can produce a graph for individual ports and use a variety of units (mega bits or mega bytes per second for example). Each graph shows the data in and out of the port, in your chosen unit.

Now, this is when using Cacti with Smokeping comes in handy. Cacti was showing that a switch disappeared at certain times of the day - the graph would simply have a gap in it. At first thought, this absense of data "clearly" indicated the switch was going offline for a period...all of them were! If you consider the chances of a regular switch shutdown across a site you'll no doubt conclude it's unlikely they all go down regularly. Looking at the smokeping graph for the same period showed that during the "outage" there was a high ping latency and 95% of all network traffic was being lost; presumably Cacti's responses were being lost in that 95%.

Cacti will also monitor disk usage over time; useful if a server suddenly goes offline due to low disk space - you can see if there was a sudden surge or if data had been growing gradually. It'll monitor whole hosts of other things too and the community produces a number of templates.



Operating System: Linux, possibly others
Tested on: Gentoo Linux, Ubuntu
License: GNU GPL
Cost: FREE

Tags: