Posts by phil

1) Message boards : Projects : Network measurements (Message 20853)
Posted 17 Oct 2008 by phil
Post:
Ok, you three start pinging around and report back to me. Looks like there won't be any need for BOINC after all ;-)

Thanks again for the replies. Hopefully I'll be able to post some results in the future.
2) Message boards : Projects : Network measurements (Message 20817)
Posted 15 Oct 2008 by phil
Post:
You got me in The Netherlands already and Dag in Canada. Between us we'd make a nice triangle. ;-)

Hehe, too bad I'm also in The Netherlands. ;-)
3) Message boards : Projects : Network measurements (Message 20815)
Posted 15 Oct 2008 by phil
Post:
Good point. Server load is definitely something to take into consideration and limiting the allowed host is a solution. Allowed hosts should not be selected on geolocation however, but that is besides the point.

Thanks for pointing me in the right direction. That config element would be useful.
4) Message boards : Projects : Network measurements (Message 20810)
Posted 15 Oct 2008 by phil
Post:
BOINC can be used for this. You could do it with a Non CPU Intensive application. (Oh look, it's also in the list there. ;-))

Hehe, yeah I have to admit I haven't had much time to look at the API yet. From the replies, I get the idea that BOINC is a very good framework to host our project. It will now be a case finding time to plan the project, do the needed research and implement it.
5) Message boards : Projects : Network measurements (Message 20807)
Posted 15 Oct 2008 by phil
Post:
Yes great, I think you all are getting the right idea. :-) Sorry for being absent, I expected to receive an email if the topic was replied to. (ok, my bad, I found the subscribe button)

@Dagorath
1 No, taking measurements will not require much CPU power, in that sense this project would not benefit from BOINC. But the the idea of having clients all over the net is very interesting as it offers a quick way of reaching different places/reaching places from different points.

2 Yes, the University of Washington is doing all kinds of measurements (Hubble, iPlane). Looking at other ways of approaching the problem of collecting useful data is a good thing, it would be better if this project could join forces with others, who knows, that may happen.

3 Mentioned already: ISPs would be interested as it could give them insight in the root of problems occurring. The Internet community would benefit. We all would benefit for that matter :-)

Just measuring from a few view points in the net will not give you a full perspective of what is going on. To get a clear view of problem areas, you would need to approach the area from different points. Another thing that is interesting is asymmetric routes in this case: one way tracerouting for instance, will not give you a full perspective of what is going on in the network.

My initial idea is that probes will not be scanning a single point in the net, they will be assigned sets of places to investigate. This list can be changed whenever a problem area is located. The rate of scans could be upped as well for a short period of time. This is where I thought BOINC would come in real handy, as it offers the possibility of assigning jobs.

Another cool thing to investigate is IPv6 adoption.

For those interested in an event that happened recently: http://www.ripe.net/news/study-youtube-hijacking.html
6) Message boards : Projects : Network measurements (Message 20787)
Posted 14 Oct 2008 by phil
Post:
The Internet is plagued by routing instabilities, reachability problems such as black holes and absurdly long latencies. For this reason I'm thinking about starting a project for network monitoring. Participants will install software that would do basic measurements using ping and traceroute, for instance. The rate of measurements will not be high: I don't want to create a net of DOS-attackers ;-)

Would it be possible to do those kind of measurements with a boinc setup?

I hope some of you can shine light on this idea for a project.



Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.