[OpenAFS] Possible explanation(s) for obvious performance problems?

Holger Rauch holger.rauch@empic.de
Sat, 24 Apr 2010 20:00:05 +0200


--huq684BweRXVnRxX
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Hi,

I'm running OpenAFS 1.4.11 (from lenny backports) on a Debian Lenny
AMD64 system (QNAP TS 809 Turbo equipped with eightSeagate 7.2k Enterprise
HDs, capacity: 1 TB each) which is a central file server (NAS),
i. e. that server runs the

- bosserver
- ptserver
- buserver
- vlserver
- volserver

daemons simultaneously. (My primary reasons for choosing OpenAFS over
Samba for SMB/CIFS, NFS v3/4, etc. was the user/rights/ACL management
which doesn't require maintaining two separate user databases
(as is the case with Samba) and the fact that OpenAFS volumes can
easily be moved to a different host (provided storage space on that
host is available).

I'm experiencing *severe* performance problems when doing a dd inside
an OpenAFS volume compared to a "plain" ext3 volume (both residing on
separate logical volumes).

Below is the dd output when dd is run inside a native ext3 volume.

=3D=3D=3D

dd if=3D/dev/zero of=3Dtestfile bs=3D1M count=3D256
256+0 records in
256+0 records out
268435456 bytes (268 MB) copied, 1,10903 s, 242 MB/s
                                                                   =20
=3D=3D=3D

Inside an OpenAFS file system, the output looks as follows:

=3D=3D=3D
                                      =20
dd if=3D/dev/zero of=3Dtestfile bs=3D1M count=3D256
256+0 records in
256+0 records out
268435456 bytes (268 MB) copied, 15,4072 s, 17,4 MB/s

=3D=3D=3D

(Both dd commands were run directly on the file server host in order
to rule out possible network latency problems as a cause for the bad
performance).

Any ideas as to where that bad performance might come from? (I do have
encryption enabled, but since it's only plain DES encryption on
current machines that most likely can't be an explanation for the
performance problem).

In addition, it's also sort of interesting to see that on Windows 7
machines the loopback adapter for OpenAFS is only set to 10 MBit, even
though the machine is phyiscally equipped with a gigbat capable NIC
and the phyiscal connection is set to 1 GBit. I noticed that after
having installed OpenAFS 1.5.72 on several Windows 7 systems. I get
transfer rates of about 8 - 10 MBytes/sec in a gigabit network from
the Windows 7 clients to the central file server. Sort of strange.

How can the speed of the loopback adapter be increased?

Thanks in advance for any hints & kind regards,

    Holger
   =20
--huq684BweRXVnRxX
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)

iEYEARECAAYFAkvTMaUACgkQbiVtWpZdKQL0DACbBuYrTAD3bJIp0rIsy8oQp88l
h/sAnRt/TDzvCN78G6ZK/Y0BruWEEkfj
=/DEZ
-----END PGP SIGNATURE-----

--huq684BweRXVnRxX--