Comments on: Internet speed decay -- We need a script!
Each month I've seen the Internet get slower and slower. The problem is not our local computers or even our internal networks. It's happening on the Internet side of our gateway/NATs... but could be just specific routes or destination servers. It's difficult to tell.
I've finally decided that I need to run a few network monitoring scripts on our servers. I think they can be quite simple, simply connecting to various other web servers (in the world of REBOL) and storing/averaging time differences. The script can be written in REBOL in less than a page. Perhaps when it connects to other servers, it could grab their timing info and store it as well.
For example, on rebol.com the time-net.r script could run via cron and read via HTTP the time-data.r file from rebol.net, rebol.org, and various other servers. It then updates its own time-data.r file. (That way, by looking at the result for one server, I can see other server summaries as well.)
I'm hoping that someone in the REBOL community has already written such a script, so I don't have to write it myself. As you know, my todo list is overflowing.
I'm sure there are many other webmasters and IT people who might use such a script as well.
So, post a comment if you have such a script! We all thank you.
10 Comments Comments:
Paul
11-Feb-2010 13:10:43 |
I'd be willing to participate if you get one. | Allen K
12-Feb-2010 14:20:37 |
Hi Carl,
Here are couple of sites that monitor the health of the internet.
http://www.internettrafficreport.com/main.htm
http://www.internetpulse.net/
| Carl Sassenrath
12-Feb-2010 18:02:15 |
Paul: ok, will let you know.
Allen: I use those... but, I need something more specific to the systems/servers we run. | Maarten
13-Feb-2010 11:55:23 |
Carl, errr.... isn't this a low priority compared to say, R3? | Carl Sassenrath
13-Feb-2010 18:30:29 |
Maarten, precisely. That's why I'm asking you to write it. I figure you especially have one in your back pocket.
It's much needed here, because we're losing up to 50% of our packets in many of our connections. We've got to determine the source so we can get it fixed.
| Maarten
14-Feb-2010 12:50:21 |
Look here under measurement tools: http://forskningsnett.uninett.no/tcpperf/
These are used by research network to track down problems (I have used iperf). As their network are state of the art, they have quite some subtle ones.
HTH | TGD
16-Feb-2010 3:08:31 |
delay: func [
{computes the response time of web-servers = delay of the internet}
host "IP-Address or host-url"
/local p rc t
] [
p: open join tcp:// host
insert p "GET / HTTP/1.0^M^/^M^/" ; HTTP GET Request
rc: copy p
close p
t: now/time/precise - now/zone ; current GMT
t - to time! fifth parse find/tail rc "Date:" none
]
example: delay "www.rebol.com:80" | Andreas
16-Feb-2010 11:30:17 |
Carl, for first diagnostics of packet loss, MTR is very helpful. WinMTR (http://winmtr.sourceforge.net/) is a version for win32. | Edoc
17-Feb-2010 13:43:42 |
I've used Charles with good results. http://www.charlesproxy.com/ Not sure how it compares to other tools mentioned here, though. | Carl Sassenrath
22-Mar-2010 21:41:58 |
Thanks for the suggestions.
Just a quick update... I've not had time to try any of the above tools but have done a few traceroutes during the problem periods... and found recently that the backbone Level3.net routers were routing the packets more than 20 hops... and doing some pretty crazy routes. This looks like a much deeper problem than I had imagined. |
Post a Comment:
You can post a comment here. Keep it on-topic.
|
