But only a few which work on linux. It's only numbers, most of it not very interesting, except for the fact, that LimeWire rules. - Maybe some of you would like to read it, though.

Perhaps one could start a serious project to collect data on gnutella's network structure...

...

Attached Files

ffl.txt (9.6 KB, 348 views)

The file is quite malformed when I read it. Do you have it in another format?

Sorry, I forgot not everyone is using less to view his text-files.

The interesting part (for me) was, that limewire's clustering obviously completely alters the balance of the messages routed between the clients, although the numbers I have from qtella can't be easily compared to the ones from LimeWire since qtella counted in- and outgoing messages and LimeWire only counted incoming messages...

Attached Files

ffs.txt (5.4 KB, 332 views)

Had to open it wordpad, but this time it's looking better. Thanks! :-)

I was amazed to see qTella's messages routed - it almost seems there's a bug in their routing mechanism (and certainly no pong caching, as you can tell the difference with LimeWire).

While LimeWire's ultrapeers feature seems to be working much better than last time I tested it personally (what a nice numbers on the bandwidth!), it does appear that gtk-gnutella is doing a better job of searching and downloading. I didn't see how many connections gtk-gnutella had at the time though - do you have that info?

Good stuff. Thanks!

-- Mike

gtk-gnutella had 8 connections.