Date: Mon, 3 May 1999 13:19:13 -0400 (EDT) From: Robert Watson <robert@cyrus.watson.org> To: Jonathan Lemon <jlemon@americantv.com> Cc: dg@root.com, advocacy@freebsd.org Subject: Re: new picture of wcarchive Message-ID: <Pine.BSF.3.96.990503131604.5183P-100000@fledge.watson.org> In-Reply-To: <199905031530.KAA09442@free.pcs>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 3 May 1999, Jonathan Lemon wrote: > In article <local.mail.freebsd-advocacy/199905021955.MAA29773@implode.root.com> you write: > > IOlite, while interesting in principle, was nothing more than a proof of > >concept and unfortunately isn't something that we would want to include in > >FreeBSD. I spoke with the researcher at length about it. For this particular > > Could you elaborate a little about that? We just had Vijay over here > to give a talk about it, and while interesting, it seemed a little limited. > In particular, the immutable buffers would seem to preclude building data > up in pieces for transmission, and they "cop'ed out" by still using normal > mbufs for small data transfers. From his description at CMU, it looked like you could easily create "aggregates" of data from files and other sources. His example to us was that each outgoing HTTP connection shipping a file might have a different header, but that we want to use the same buffers where possible for the file itself. Not having looked at their code, I'm only going on what he said, but it sounded like they had a flexible API for manipulating aggregates of these immutable buffers, which handles a lot of the cases we're concerned about. He didn't mention the copy out for using mbufs to us, but given the page-wise behavior they describe, that makes sense (i.e., the use of virtual memory page permissions to guarantee immutability of sets of buffers stored in those pages). Their performance results were very impressive. Does sendfile() use the same socket buffer storage for all copies of an outgoing file? Robert N Watson robert@fledge.watson.org http://www.watson.org/~robert/ PGP key fingerprint: AF B5 5F FF A6 4A 79 37 ED 5F 55 E9 58 04 6A B1 Carnegie Mellon University http://www.cmu.edu/ TIS Labs at Network Associates, Inc. http://www.tis.com/ Safeport Network Services http://www.safeport.com/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-advocacy" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.96.990503131604.5183P-100000>