Messages in this thread | | | Date | Wed, 2 Jun 1999 23:49:17 -0500 (CDT) | From | Oliver Xymoron <> | Subject | Re: zero-copy TCP fileserving |
| |
On Thu, 3 Jun 1999, Johannes Erdfelt wrote:
> > This simple zero-copy isn't really a win on most PC hardware as it takes > > about the same amount of time to do a checksum as it does to do a checksum > > and copy. > > What about the new network cards which can do checksums in hardware?
That's currently a special case. The architecture is optimized for the common case and currently compares quite favorably to other systems, even those that use zero-copy approaches..
> > A more powerful scheme is being investigated by sct and a couple others. A > > related scheme worked out on FreeBSD and presented at a recent Usenix > > could cache checksums, for example. > > Hmm, caching checksums? I wonder how that would work. It would only work > on data which is somewhat repeated. This would probably be a no win for > encrypted data. Albeit, I haven't read the paper yet, so I dunno how it > works.
The Linux network stack is much faster than any decent software encryption scheme, all of which make a dozen or more read/write rounds on its input, so that isn't a very interesting case. The interesting cases for networking are file and web serving, and these seem fairly amenable to the IO-Lite style approach (even some dynamic content).
-- "Love the dolphins," she advised him. "Write by W.A.S.T.E.."
- 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/
| |