Internet speed monitoring?

[mrbios]

Are there any programs around that can monitor multiple internet connections and monitor the speed of the connections even through a proxy?

We keep getting one teacher here constantly moaning about the internet being slow when it isn't it would be nice to be able to show him a monitored connection showing that it is NOT slow he's just being impatient/the page hes trying to load is on crap hosting (75% of the time it's the VLE which is on uniservity)

Anything like that around?

[OutToLunch]

Do you have snmp access to your router? I used prtg here to track the in/out rate at the edge of the network and show that although it might seem slow to an individual the whole connection was being maxed out across the school...

[tom_newton]

Try this. Runs on linux, needs perl, wget.

Hope someone finds it useful - just spent 10 mins updating it to work with a modernish wget.

It downloads a list of sites, wioth and without caching. Then tells you how long, in seconds, that took.

Cron it in to run every 5 mins, then get excel out and hack up a tasty graph.


Will need editing to suit your proxy and site list.

[OutToLunch]

Try this. Runs on linux, needs perl, wget.

my @sites = ("http://www.google.co.uk","http://news.bbc.co.uk","http://www.cnn.com","http://blog.flickr.net","http://playboy.com");

Possibly a good idea not to run with the last default site Like the theory about 24 hour downloads too

[tom_newton]

Heh, the last site is in there as a "lets see the response time of a blocked site" as it's likely to be blocked by *any* filter. This is a dirty hack to work out if it is the filter that's tardy in responding, or if the internet connection is choked to death.

[OutToLunch]

Ahh so not "I'll just wget -r me playboy.com for the weekend"

[tom_newton]

Nah.. anyway it only gets the first page + all images.

Irritatingly, playboy's frontpage now contains pictures of boobies. It never used to (it was provocative, but not porn). This made it a good "test site" because if the filter wasn't on/was broken, at least you wouldn't have the embarrassment of "screen-o-tits". Bah.

[User3204]

Send them to Speedtest.net - The Global Broadband Speed Test
If you have a decent cache it will be stupidly fast, we're getting about 60mb downloads through a 10mb connection... otherwise I would check via Linux friendly ISP

[ZeroHour]

Do you have snmp access to your router? I used prtg here to track the in/out rate at the edge of the network and show that although it might seem slow to an individual the whole connection was being maxed out across the school...

That looks like a pretty nice bit of software. Do you need it on a server as such or did you just install it on your client?
Are you using free or have you purchased?

[ICTNUT]

@zerohour: this is what you need PRTG Network Monitor - intuitive network monitoring software the freeware version allows for 5 sensors to be checked.

[ZeroHour]

Hey Oz,
Its more of a case of if I install it on my client will it beat down my performance?

[ICTNUT]

I use the latest version on a Quad-Core 4Gb ram PC and I dont really see that much of a hit, but then I am only monitoring the Cisco gateway nothing really that intensive.

[pete]

You should be able to phone your RBC and ask for a mrtg snapshot of your connection. Some (most?) log the usage by default, though if you want a detailed "between here and here" snapshot it might take longer.

Nah.. anyway it only gets the first page + all images.

Irritatingly, playboy's frontpage now contains pictures of boobies. It never used to (it was provocative, but not porn). This made it a good "test site" because if the filter wasn't on/was broken, at least you wouldn't have the embarrassment of "screen-o-tits". Bah.

Use 2600.org instead?

[Michael]

We keep getting one teacher here constantly moaning about the internet being slow

Generally speaking the best way to demonstrate it's not your connection but the host of a website is to use a broadband speed tester. The BBC have one which is pretty good and seems fairly accurate. Running that in front of the user should prove your point.