[OpenAFS] Performance

[Neulinger, Nathan]

The user space daemon in afs doesn't do much of anything. It's there
primarily to launch kernel threads and set up the cache.=20

-- Nathan

------------------------------------------------------------
Nathan Neulinger                       EMail:  nneul@umr.edu
University of Missouri - Rolla         Phone: (573) 341-4841
Computing Services                       Fax: (573) 341-4216


> -----Original Message-----
> From: Nathan Ward [mailto:nward@esphion.com]=20
> Sent: Thursday, February 27, 2003 1:52 PM
> To: emoy@apple.com
> Cc: Michael Robokoff; openafs
> Subject: Re: [OpenAFS] Performance
>=20
>=20
> Well I'd expect that it goes slower as your cache size is=20
> exceeded as it=20
> then needs to start getting that data to the server. Or is=20
> the cache for=20
> read operations only?
>=20
> I notice that there are around about the same number of=20
> packets/sec as=20
> context switches/sec on my client machines. I wonder if=20
> switches between=20
> userland and kernel could be to blame... ? Who sends packets=20
> in OpenAFS,=20
> the userspace daemon or the kernel?
>=20
> Nathan
>=20
> emoy@apple.com wrote:
>=20
> > Could the slowness you see with your dd write test be=20
> related to the =20
> > cache exhaustion issue that I raised recently, when writing a file =20
> > larger than your cache size.  Your test writes a 1 GB file, so if=20
> > your  cache is smaller than this, you will see poor=20
> performance once=20
> > your  cache size is exceeded.
> >=20
> --------------------------------------------------------------
> ----------=20
> > --=20
> > Edward Moy
> > Apple Computer, Inc.
> > emoy@apple.com
> >
> > (This message is from me as a reader of this list, and not=20
> a statement
> > from Apple.)
> >
> > On Thursday, February 27, 2003, at 11:15  AM, Nathan Ward wrote:
> >
> >> I see pretty bad performance to tell you the truth.
> >> I can read and write ~60mb/s directly to my raid array, but when=20
> >> using  OpenAFS (locally or remotely) to the same array, I=20
> get around =20
> >> 6-10MB/s, I have seen up to 25MB/s over a peice of=20
> 1000Mbps fibre. =20
> >> Client and Server are both dual P3-1ghz with 1024mb ram. I notice=20
> >> the  context switches on the server at this time jump to ~10000/s,=20
> >> and on  the client ~40000/s. I imagine this is the source of my=20
> >> slowdown, but  I havn't had a chance to look into it.
> >>
> >> I'd be interested if anyone else has the same level of context =20
> >> switches going on.
> >>
> >> This is while doing a large sequential write operation (dd =20
> >> if=3D/dev/zero of=3D/afs/alb-nz.esphion.com/public/dd.out bs=3D256k =
=20
> >> count=3D4096).
> >>
> >> Michael Robokoff wrote:
> >>
> >>> Does anyone have any Open AFS performance information they can=20
> >>> share  with me. I plan on doing a couple benchmarks and I=20
> would like=20
> >>> to have  some idea of what to expect.
> >>
> >
> > _______________________________________________
> > OpenAFS-info mailing list
> > OpenAFS-info@openafs.org
> > https://lists.openafs.org/mailman/listinfo/openafs-info
> >
> >
>=20
>=20
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info
>=20