Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have a CentOS machine that is being used as a PBX. One interface is going out to the local network and the other interface is currently (pending equipment arrival) plugged directly into a Metro Ethernet connection from our provider.

I would like to monitor the bandwidth going out to the internet (essentially everything entering or exiting eth1) from within the box itself, is there a very simple tool that can give me active bandwidth usage per interface?

Like that I would be able to tweak codecs and other things to get to an optimal call limit.

(any other comments about such a setup would be appreciated:) )

Thanks.

share|improve this question
    
On the first day that I switched the company over from using VOIP over DSL to VOIP over a 'T1' class connection, the call quality suddenly dropped, I kinda forgot that even DSL is still a far bigger pipe than a 3.0 MB dedicated pipe. –  yayim Oct 8 '09 at 0:40

4 Answers 4

up vote 1 down vote accepted

There are a few simple text-based tools that you can use:

  • bwm-ng shows numerical counters for traffic rate on every interface (rx, tx and total) at a configurable refresh rate (default of 500ms)
  • iftop shows numerical and bar graph indicators of traffic based on source/destination hosts
  • iptraf shows numeric packet and byte counts based on connections (source/dest host and port) or packet counts per interface, port or physical address.

(all should be available for common distros, but maybe not in the default repos)

In my personal experience, if call quality is the problem, you should look more into latency and jitter than bandwidth. Asterisk (if that's what you're using) has some nice builtin tools and debugging, and some other network tools can help.

If you want to know bandwidth, it would probably be better to use something like Cacti or Munin or something else based on RRDtool to graph the statistics over time.

share|improve this answer
    
vnStat looks good, and I also remember using ntop (it isn't installed on my router now, but I remember using something on another box that I liked the best, perhaps ntop was it). –  Jason Antman Oct 8 '09 at 3:21
    
I use vnstat and love it. There is also a GUI or web interface for it, although I prefer CLI. Its simple, easy, and free. –  cop1152 Oct 8 '09 at 3:47
    
I dont think it's latency and jitter, because based on what I have seen so far (and I will definitely be looking at the builtin tools) with pinging, i am getting consistent under 5ms pings to the VOIP provider, variations being in the tune of 1 ms, so I would doubt it's that (Problem is, I did not check this during business hours...). We also have a dedicated line with SLA for latency and jitter. Vnstat looks to do the trick for now (seems to have a history recall as well). Thanks for the suggestions, I will look in to this. –  yayim Oct 8 '09 at 6:15

I've used 'iftop' before, I'm not sure if that's exactly what you're looking for but there seems to be a package for it here: http://dag.wieers.com/rpm/packages/iftop/

share|improve this answer
    
Thanks for the suggestion, it looks to be pretty much exactly what I needed (as some of the other suggestions here), but the repo you listed (and several others I tried) were down...I went with vnstats, and it did the trick. –  yayim Oct 8 '09 at 6:00

You can try vnStats of bandwidthd

share|improve this answer
    
Went with vnstats (as of yet I dont need the full graphing thing, just some live interface data.) Thanks for the suggestion. –  yayim Oct 8 '09 at 6:02

I've just started using VoipSpear. It's simple, free.

Here is the link http://www.voipspear.com

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.