I would like to do an automatic audit of bandwidth of my remote offices. I understand that there are many websites where I could ask each user to go to a webpage and run the test. I was hoping there was some software that I could run on our web server and a script that I could send out that would run the test and record the results to a central web interface. The remote sites are all running some form of Windows and the server environment can be Linux or Windows based.


You should really be doing this by having SNMP monitoring on the border router(s) of your remote sites. Tools like MRTG are made specifically for this. Except for your monitoring server, this has nothing to do with the servers at the remote sites. This is network monitoring, and should be focused on the network devices themselves.

  • 1
    I'd just like to add that something like Cacti or Zenoss might be a more polished product to use than the base MRTG package. – Aaron Copley Jan 6 '11 at 19:09
  • Sure sure; last time I was in charge of such things (2004), it was with MRTG. – mfinni Jan 6 '11 at 19:51
  • 1
    Those "SpeedTest.net" type sites don't really give the type of information that an admin needs for thorough investigation. Border monitoring FTW! – Wesley Jan 6 '11 at 20:03
  • @Wesley : That too. I didn't want to bring it up, but you're entirely right. – mfinni Jan 6 '11 at 20:04
  • Another thing to note here is that we do not have enterprise firewalls or VPNs at the remote sites. So SNMP is really not an option. Many are shared networks amongst other orgs. – keithosu Jan 14 '11 at 4:26

I agree with mfinni's idea but if you absolutely feel the need to test your bandwidth you can easily create a script to use wget to download one of those large test files from a test site. wget's output will show what you downloaded, how long it took and the average speed for the download. If necessary run the test on a schedule.

  • That can be useful additional information, but it's hardly an audit of used/available bandwidth for a given site. – mfinni Jan 6 '11 at 21:33
  • @mfinni, I fully agree but the way I read the question (perhaps incorrectly) the OP wants to test the available bandwidth, rather than that being used. MRTG and the like cannot determine true available bandwidth. At best they can read what the connected devices are capable of. e.g. Having a 1000Mb NIC doesn't mean you have 100Mb available bandwidth. – John Gardeniers Jan 6 '11 at 22:21
  • That's why, on MRTG, you can set the actual limit of a line. If you have a 10 MBit Eth0 interface that goes to a CSU/DSU on a T1, you know it's a 1.544 MBit line. If you're using 1.044 MBit, you know you have .5 MBit available. – mfinni Jan 7 '11 at 1:53
  • @mfinni, I'm well aware of the MRTG settings but normally the purpose of testing bandwidth is to find out whether or not you can actually achieve what you believe it should be. Just because an ISP says you have X bandwidth (and it's always nominal anyway) doesn't mean you can really achieve that. Hence the need to test. – John Gardeniers Jan 7 '11 at 2:42
  • I suppose if you're talking about consumer-grade internet links, that could be valid. Of course, you could also generate load using something like netcat between nodes on different sides of a link and see if that can actually fill up your line. Depending on a free, potentially unreliable, service like those online bandwidth tests is a pretty coarse-grained tool. – mfinni Jan 7 '11 at 3:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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