Da-Chief
03-25-2008, 07:34
http://i.dslr.net/urls/84/71184.gif (http://www.dslreports.com/shownews/Once-Again-Someone-Proposes-Fixing-TCP-92958)
Several users have sent in this ZDNet discussion (http://blogs.zdnet.com/Ou/?p=1078) about reworking of the TCP (Transmission Control Protocol) congestion control. In short, author George Ou, who frequently rants against the idea of network neutrality, suggests that networks can fix P2P users' uneven consumption of bandwidth by fixing the existing TCP implementation that uses Van Jacobson s 20+ year old AIMD algorithm.Under Jacobson s algorithm, TCP currently gives the user with 11 opened TCP streams 11 times more bandwidth than the user who only uses one TCP stream. Under a weighted TCP implementation, both users get the same amount of bandwidth regardless of how many TCP streams each user opens. This is accomplished by the single-stream application tagging its TCP stream at a higher weight than a multi-stream application.
While Ou (who has mis characterized (http://www.corpsman.com/shownews/90036) the net neutrality debate badly in the past) buries a number of not so subtle net-neutrality barbs in his piece, the discussion is still an interesting one. Still, we've seen countless people propose a "fix" to TCP over the years, and it never happens -- in part because the protocol ultimately works pretty well. Some additional discussion over in the Slashdot (http://tech.slashdot.org/tech/08/03/24/1359212.shtml) comment section.
read comment(s) (http://www.dslreports.com/shownews/Once-Again-Someone-Proposes-Fixing-TCP-92958)
More...
Several users have sent in this ZDNet discussion (http://blogs.zdnet.com/Ou/?p=1078) about reworking of the TCP (Transmission Control Protocol) congestion control. In short, author George Ou, who frequently rants against the idea of network neutrality, suggests that networks can fix P2P users' uneven consumption of bandwidth by fixing the existing TCP implementation that uses Van Jacobson s 20+ year old AIMD algorithm.Under Jacobson s algorithm, TCP currently gives the user with 11 opened TCP streams 11 times more bandwidth than the user who only uses one TCP stream. Under a weighted TCP implementation, both users get the same amount of bandwidth regardless of how many TCP streams each user opens. This is accomplished by the single-stream application tagging its TCP stream at a higher weight than a multi-stream application.
While Ou (who has mis characterized (http://www.corpsman.com/shownews/90036) the net neutrality debate badly in the past) buries a number of not so subtle net-neutrality barbs in his piece, the discussion is still an interesting one. Still, we've seen countless people propose a "fix" to TCP over the years, and it never happens -- in part because the protocol ultimately works pretty well. Some additional discussion over in the Slashdot (http://tech.slashdot.org/tech/08/03/24/1359212.shtml) comment section.
read comment(s) (http://www.dslreports.com/shownews/Once-Again-Someone-Proposes-Fixing-TCP-92958)
More...