Hi, We ran some testing on caching here at limepeer. Our results and comments are available are available at: http://www.limewire.com/developer/GDFDoc.html ...
486
Vinnie
info@...
Mar 1, 2001 9:54 pm
... I think that any cache implementation should be designed to benefit only the local user. In other words, cached results should not be sent out by one...
487
Vinnie
info@...
Mar 2, 2001 12:00 am
Instead of this passive network crawling model, why don't we just have a 'push' model where each servent reports its IP address and all of its neighbors every...
488
Rosser Schwarz
rosser@...
Mar 2, 2001 12:11 am
... not in a million years. a set of decentralised servers that joe user can host, absolutely. -they- can publish their ips to the network every five minutes....
489
Vinnie
info@...
Mar 2, 2001 12:34 am
... Well it is centralized already. I was just thinking about how to apply a client/server model to make it incredibly more efficient. In a worst case...
490
Rosser Schwarz
rosser@...
Mar 2, 2001 12:43 am
... in a worst case scenario, where, having no one to sue the pants off, certain parties who shall remain unnamed will be able to take down a central server,...
491
Rosser Schwarz
rosser@...
Mar 2, 2001 12:50 am
... it would even be easy to write a node that, upon not receiving the "i'm caching ips and soliciting ips for caching" message from the cache--a message which...
492
Serguei Osokine
osokin@...
Mar 2, 2001 2:34 am
... I dunno... we are starting to talk about the network self- organization in both cases presented here. Ben's been pushing that for a long time and in...
493
Rosser Schwarz
rosser@...
Mar 2, 2001 5:10 am
... [...] ... not to nit-pick, but to my mind the original idea was more of imposing an order on the network than building the capability to self-organise. to...
494
Hagelslag user
hg@...
Mar 2, 2001 8:07 am
... Unfortunately I haven't been on this list long enough to understand what you mean. Does this have somethink to do with the four-spaces-indexing thing or...
495
KOBA
koba@...
Mar 2, 2001 8:39 am
Hi, there! I will address some thoughts in terms of Japanese gnutella user and so on. [For Gnutella users using Double Bytes Character Code] Now, Gnutella...
496
steve bryan
steve_bryan@...
Mar 2, 2001 11:09 pm
There was actually a useful suggestion in a slashdot discussion today (specifically the one about the gnutella "worm"). The weakness here is to an adversary...
497
Serguei Osokine
osokin@...
Mar 3, 2001 12:03 am
... Here is how I would beat it if I'd be writing a virus: I would answer all the queries, but would masquerade the response as the one coming from the...
498
Ben Houston
ben@...
Mar 3, 2001 12:07 am
... Curiosity sometimes kills the cat. ;-) ... It will catch this virus but I'm not really sure if besides that point it is that useful. This type of...
499
Tony Kimball
alk@...
Mar 3, 2001 12:45 am
The first thing I would do to reduce spam and virus impact is put the sha-1 hash of a query_hit's content in the location, in lower-case hexadecimal, so that...
500
Lucas Gonze
lucas@...
Mar 3, 2001 12:52 am
A fairly trivial short-term solution is to call virus scanning software from your servent on downloading an executable file. I should point out the...
501
Serguei Osokine
osokin@...
Mar 3, 2001 2:15 am
... Hey, from the purely technical standpoint it is not a short-term solution - it is at least a medium-term (until we overload McAfee and friends :-), and...
502
Lucas Gonze
lucas@...
Mar 3, 2001 3:57 am
... Maybe don't even try to do it without purchasing an anti-virus package. Let people who already have anti-virus software use it (there's gotta be some kind...
503
steve bryan
steve_bryan@...
Mar 3, 2001 7:04 am
... There is a cartoon at the beginning of a classic book about computer security and firewalls that shows various monsters lining up each to have his height...
504
Iain Wade
iwade@...
Mar 3, 2001 9:00 am
The virus writer would only have to randomly pad their .exe with NOOP or other meaningless instructions to effectively change their signature .. and, assuming...
505
Lucas Gonze
lucas@...
Mar 3, 2001 8:57 pm
I have heard that the typical number of nodes on a napster server is 4-5 K. Can anyone confirm this or cite a source?...
506
Greg Bildson
gbildson@...
Mar 3, 2001 9:14 pm
I heard the max is 10,000 a lot and then recently someone said 16,000. I still believe that it is 10,000 as some old news articles said. As to the average on...
507
Lucas Gonze
lucas@...
Mar 3, 2001 10:45 pm
Question: which servents gather IPs by sending pings instead of using the limewire or Clip2 IP caches?...
508
Greg Bildson
gbildson@...
Mar 3, 2001 10:58 pm
I think they all do both. LimeWire will go back to a host cache after a half hour to get better pongs if needed. Most will start by connecting to a host...
509
Lucas Gonze
lucas@...
Mar 3, 2001 11:29 pm
Does LimeWire not send pings unless IPs from the host cache didn't generate enough useful connections? I'm trying to figure out why there are so many pings...
510
Greg Bildson
gbildson@...
Mar 3, 2001 11:40 pm
Pretty much every client pings a connection after opening it and from the other side. BearShare has done some throttling of ping/pong traffic with 1 per...
511
Lucas Gonze
lucas@...
Mar 3, 2001 11:50 pm
So you are sending pings but not using the pongs? If that is the case, not sending unnecessary pings seems like an easy way to reduce pongs. Am I missing...
512
Andrea Denzler
list@...
Mar 4, 2001 2:55 pm
I am also very interested in statistics about users connected to Napster and Gnutella. Does anyone know where to get these info's? How many average users are...
513
Ben Houston
ben@...
Mar 4, 2001 5:57 pm
Hi All, For those that love to talk about P2P non-stop here is another fix for you. This list is trying to maintain an elite status though. Justin Chapweske ...
514
John Marshall
swabby@...
Mar 5, 2001 4:00 am
I propose extending the size of the public payload byte of the QHD to 2. Meaning 2 bytes of information with each query hit returned to a client. The first...
