Re: [squid-users] diskd performs badly

You misremember Edward,

DiskD has never topped aufs in any tests I have done or seen results for
on Linux.

The margin isn't huge, though. aufs is CPU-hungry compared to diskd, so
if you've got a lot more disks than CPU horsepower, DiskD might be
preferable (but I wouldn't bet on it). Sometimes the balance of
hardware can lead to surprising results...Duane found, for example, that
ReiserFS is slower on his test box than ext2. I haven't really figured
that one out yet, as I've never gotten those results, but I reckon it is
due to the balance of available hardware resources. His box had several
10k RPM SCSI disks, and a relatively slow CPU, whereas my test boxes are
usually 1 or 2 slower IDE disks and a similar or faster CPU.

Edward D. Millington wrote:

> Henrik!
>
> Didn't Joe Cooper ran some test comparing DiskD with Aufs and so on,
> with DD coming slightly on top?
>
> I have personally find, without testing, just "feeling", a slight
> performance when using DD. I have been using DD for over 2 years now.
>
> Edward

-- 
Joe Cooper <joe@swelltech.com>
Web caching appliances and support.
http://www.swelltech.com