MRTG on multiple routers

mzinzmzinz Member Posts: 328
I'm getting some interface usage monitoring going on several of our routers. I want something that can monitor all of our WAN links reliably and update on its own.

I started using MRTG. The only issue is that we have ~100 routers to monitor, and every single mrtg process (perl, actually), takes up 10-15mb of ram. This means that I'm going to be using between 1 - 1.5gb of ram just to monitor all of these devices.

Are there any other options? I like the graphs it makes, but it is still a hassle to setup (I created a HTML page where I just IMG SRC to the graph I want for each router... ie: end result would be a simple HTML page with 100 graphs...
_______LAB________
2x 2950
2x 3550
2x 2650XM
2x 3640
1x 2801

Comments

  • tmlerdaltmlerdal Member Posts: 80 ■■□□□□□□□□
    I've been using MRTG for a long time now, and one thing I have done is just setup to monitor the head end routers. Seemed pointless to me to monitor the far end router passing that traffic back across the link and basically having the same graph twice.

    Of course, this is only if you are wanting to gather circuit stats and not other info like CPU utilization, etc.

    Another thing you could do is adjust your polling interval instead of every 5 minutes, bump it out farther and see if that helps.

    I'm running MRTG on an older server class machine and have tried to automate it as best I can with some Perl scripts for generating the webpages.

    If you have the $$, you could look at the products from Solarwinds. I know some guys that have used it in the past and love it.
  • mzinzmzinz Member Posts: 328
    tmlerdal wrote: »
    I've been using MRTG for a long time now, and one thing I have done is just setup to monitor the head end routers. Seemed pointless to me to monitor the far end router passing that traffic back across the link and basically having the same graph twice.

    Of course, this is only if you are wanting to gather circuit stats and not other info like CPU utilization, etc.

    Another thing you could do is adjust your polling interval instead of every 5 minutes, bump it out farther and see if that helps.

    I'm running MRTG on an older server class machine and have tried to automate it as best I can with some Perl scripts for generating the webpages.

    If you have the $$, you could look at the products from Solarwinds. I know some guys that have used it in the past and love it.

    Thanks for the post. Solarwinds seems nice, but I was worried that I would be paying thousands for something that is way more advanced than what I need.

    Monitoring the head-end with MRTG would be GREAT... unfortunately the T1s don't terminate on my router directly (MPLS cloud).

    Thanks though :)
    _______LAB________
    2x 2950
    2x 3550
    2x 2650XM
    2x 3640
    1x 2801
  • Forsaken_GAForsaken_GA Member Posts: 4,024
    Look into Cacti.
  • SysAdmin4066SysAdmin4066 Member Posts: 443
    Solarwinds is very nice, our provider (another agency in our org) uses it for their WAN. Also take a look at Nagios, its open source monitoring and configuration (snmp) sofware that is very powerful.
    In Progress: CCIE R&S Written Scheduled July 17th (Tentative)

    Next Up: CCIE R&S Lab
  • mzinzmzinz Member Posts: 328
    Thanks for all the replies, very helpful.

    I ended up setting up a single instance of MRTG with multiple routers inside. This way it uses only a few mb of RAM and still monitors almost 100 routers.

    I then created a simple HTML page which has <IMG SRC=> to the PPP interface on each router... this way I can view ALL T1 connections simply by going to one page.

    Thanks again for all the help!
    _______LAB________
    2x 2950
    2x 3550
    2x 2650XM
    2x 3640
    1x 2801
Sign In or Register to comment.