Messages in this thread | | | Date | Thu, 3 Jun 1999 22:24:33 +0200 | From | Matthew Wilcox <> | Subject | Re: XTP: A better TCP than TCP |
| |
On Thu, Jun 03, 1999 at 01:49:30PM -0400, Greg Lindahl wrote: > > I think a much more interesting protocol to have in the kernel would > > be Bell Labs' IL. It's implemented in Plan 9, and they say they are > > very happy with its performance. It's a sequenced packet protocol, > > ideal for NFS, CORBA messages and pretty much anything which currently > > either implements its own out-of-order + retransmit strategy over UDP > > or puts packet boundaries into TCP. > > Note that IL has no flow control. You may not think you need flow > control, but the minute you have several people talking to 1 at high > speed, life gets interesting. A group I used to work for (Legion) had > its own IL-like protocol, and we finally decided that TCP's flow > control was a huge win locally (our Legion-MPI programs were > overrunning the 64k OS buffers), and TCP's adaptive stuff was far > superior over WANs.
As I understand it, the idea behind not requiring flow-control in IL was that the higher level protocols would take care of this; in 9P (which is the main user of IL), there's a reply to each request, so if the server is replying slowly, the client will slow down to compensate. I don't think IL is suitable for simply sending packets to hosts, it must have a request-response protocol layered on top of it.
Yes, this requires the higher level protocol knows about doing flow-control stuff like nagle, but the idea is that the TCP layer doesn't _really_ know what the application requires and setting options such as TCP_NODELAY is a crude way of telling it.
-- Matthew Wilcox <willy@bofh.ai> "Windows and MacOS are products, contrived by engineers in the service of specific companies. Unix, by contrast, is not so much a product as it is a painstakingly compiled oral history of the hacker subculture." - N Stephenson
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |