Heres the link: http://netalyzr.icsi.berkeley.edu/
Having looked through the logs that generates when used on my connection, I can see it as a useful debugging tool. It is certainly comprehensive - everything from bandwidth, latency and jitter testing to a comprehensive set of protocol tests. However, it is a little limited by the server being in the US (so there's a lot of hops between you and the server - not ideal for testing bandwidth, latency and jitter - my 80/20 FTTC connection tests at 13/10 according to the log), and there will be a great deal of 'noise' where it brings up things that really aren't an issue.
The only thing it spots on my connection is the firewall dropping all incoming and outgoing Internet traffic on TCP ports 135 and 139 and UDP ports 137 and 138 traffic. These protocols should never travel over the Internet unless wrapped in a VPN to a trusted destination.
The clocks on all my machines are set via NTP (or W32time on the Windows boxes) and I run a local stratum 1 NTP server against GPS, so the clocks on all my servers vary between typical figures of <50us from UTC (the stratum 1 server), <0.5ms from UTC (everything running NTP) and <0.8s from UTC (those boxes running W32time - it could do so much better in disciplining the clock <sigh>). As already explained, the DNS server on my router supports DNSSEC, though this is of limited use at present.
For most people, having a local clock that's a few seconds out or having no DNSSEC support is of no consequence.
This test suite will spot useful things such as broken path MTU discovery (usually by overzealous blocking of ICMP). However, the protocol tests appear to be based purely on passing a small amount of traffic over the well-known port, which may not give a true picture of whether that protocol will actually work over that connection (the test may not trigger the likes of a Snort rule).



Pages in this thread:
Print Thread
deleted