Hi,
I've been busy doing a couple of stress tests lately. A disturbing finding
is that the 1.4.7 fileserver achieves nearly 25% higher data throughput
than 1.4.8 and 1.4.10 respectively.
My test server is amd64_rhel50 with kernel 2.6.18-92.1.22.el5 and uses two
bonded NICs.
All test clients are amd64_rhel50 as well, 2.6.18-128.1.6.el5, using 64MB
memcaches.
The test comprised a simultaneous "dd if=/afs/... of=/dev/null".
To 5 and more clients, a 1.4.7 server will consistently serve ~160MB/s,
whereas higher versions yield <130MB/s.
All tests were performed using client version 1.4.7 (with RxOSD patch) and
1.4.10 (with and without RxOSD patch) respectively. Server versions tested
were 1.4.7, 1.4.8 and 1.4.10 (in addition to some with RxOSD patch).
The server was built with --enable-bitmap-later
--enable-bos-restricted-mode --enable-bos-new-config --enable-fast-restart
--enable-supergroups --enable-largefile-fileserver --enable-transarc-paths
--with-krb5-conf
It's invoked with -L -p 50 -rxpck 300 -s 1000 -l 1000 -cb 65535 -b 250 -syslog
Can anyone confirm this?
Regards
Felix
_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel
