Wow, thanks for the responses. I'll keep working through them, but speedtest mini looks promising since it'll be familiar with the shaky speedometer.
Summary: - *Speedtest mini* - I think Ookla is the commercial version, with customisation and branding and judging by James Bensley's comments below, they auto-expire the software so you have to take the Ookla path eventually or script your way out of it. - *Smokeping*'s behaviour can probably be duplicated by RPM probes and thresholds in the SRX CEs, though it might produce better looking graphs. - *ICSI Netalyzr* looks like a good all-purpose tool. Certainly something worth testing, maybe worth giving to field engineers, definitely worth using during intial operational readiness testing. - *Samknows*' devices might be overkill, I don't think there's a great budget for this, plus it's a private network so we can't offer stats - *NDT* - might be unpopular internally due to the words "patch" and "kernel". I'd need to see a demo, but it looks like it'd take a while to get up and running. Thanks again, this is what I hoped UKNOF mailing list would be about and you've shown how useful it can be. I'll report back on any further findings. Ben On 5 December 2011 11:04, Ben Ward <[email protected]> wrote: > Hi NOFfers, > > Despite distrusting HTTP Speed Tests I'm investigating speed test > software/hardware which I can deploy inside a private network > infrastructure to prove problems off our network when only HTTP is > available to diagnose "slow internet". There are several customer proxy > servers which we suspect introduce problems, so we're trying to take them > out of the picture. > > Does anyone have any recommendations here? We are looking for something > convincing enough for end-users, but transparent enough for us to make > sense of the results. Also, we'd like to log the results so we can monitor > performance. > > Ben > > -- > [email protected] | http://crouchingbadger.com > -- [email protected] | http://crouchingbadger.com
