Messages in this thread | | | Subject | Re: Kernel interface changes (was Re: cdrecord problems on | Date | Mon, 08 Feb 1999 19:46:30 -0500 | From | "Brandon S. Allbery KF8NH" <> |
| |
In message <sjmyam8bgvw.fsf@datkins.ihtfp.org>, Derek Atkins writes: +----- | "slow", even when it's in the kernel. I don't even want to think | about how gawdawful slow it would be elsewhere (I've never used Arla, | so I don't know exactly how slow that is). I do believe that the +--->8
Arla's not too terrible; file accesses are intercepted and exchanges with the AFS servers are done if necessary to get the file into the cache, from that point on it's all local processing redirected to the cache inode until the file is closed. (Arla caches by files, not by blocks, which has the advantage of speed while the file is open but the disadvantage that the entire file has to be shuttled between the server and the client on open/close.)
The biggest speed hit in this proposal is the double network access, between Linux and the translation server and between the translation server and the AFS server. The AFS-NFS translator is anything but fast, in my testing, even when hosted on a box with fast networking.
-- brandon s. allbery [os/2][linux][solaris][japh] allbery@kf8nh.apk.net system administrator [WAY too many hats] allbery@ece.cmu.edu carnegie mellon / electrical and computer engineering KF8NH We are Linux. Resistance is an indication that you missed the point.
- 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/
| |