- From: Ashish Kolli <ashishk@wankel.me.cmu.edu>
- Date: Wed, 21 Feb 1996 08:19:52 -0800
- To: BearHeart / Bill Weinman <bearheart@bearnet.com>
- Cc: www-talk@w3.org
Hi, I think what Rob was trying to get at is the bandwidth issue: How well would the HTTP-NG protocol perform in a congested network when compared to the HTTP1.x . Since the overhead of establishing a new connection for each request is eliminated in HTTP-NG we would expect a certain amount of performance benefit. However when one is connected by computers that are "quite distant" the TCP/IP protocol might introduce some latencies in terms of packet loss and retransmission for a connection that is open for a long period of time, because the load from other factors like "Sprintlink" might vary drastically. - Ashish On Feb 21, 11:35am, BearHeart / Bill Weinman wrote: > Subject: Re: HTTP-NG Server & Client Mechanisms > At 11:32 pm 2/20/96 MST, Rob Hartill spake: > >There are far too many reports which tell us how well a client talks > >to a server that is only a few feet away... not much help unless > >we know how it behaves when fighting against network black holes such > >as Sprintlink. > > I'm confused. What does this have to do with client performance? > > What can a client do about some unknown piece of hardware between it > and its server? > > > +--------------------------------------------------------------------------+ > | BearHeart / Bill Weinman | BearHeart@bearnet.com | http://www.bearnet.com/ > | Author of The CGI Book -- http://www.bearnet.com/cgibook/ >-- End of excerpt from BearHeart / Bill Weinman -- ------------------------------------------------------------------------------ Ashish Kolli, Carnegie Mellon University Email : ashishk+@cmu.edu WWW : http://www.cs.cmu.edu/~ashishk/ ------------------------------------------------------------------------------
Received on Wednesday, 21 February 1996 12:51:07 UTC