On Tue, 30 Sep 2008, Cosimo Streppone wrote:
In data 30 settembre 2008 alle ore 00:09:52, James Smith <[EMAIL PROTECTED]>
ha scritto:
On Mon, 29 Sep 2008, Cosimo Streppone wrote:
In data 29 settembre 2008 alle ore 23:45:05, James Smith
<[EMAIL PROTECTED]> ha scritto:
There are good reasons to store images (especially small ones) in
databases (and with careful management of headers in your mod_perl).
If you have "proper" metadata, you can go and delete your files.
We have "proper" meta data
Yes, sorry. I was thinking about our case.
we produce and delete anwhere up to and including 1/2
million files per day - and
the deletion is the crippling stage on a journalled file system
I see. Again, our case is very different though.
99,99% is create/add/modify and we almost never delete.
What filesystem/os do you use?
It has to be a shared filesystem - so at the moment GPFS/red had; we are
moving over to a memcached system to store and server the temporary images;
otherwise we would be stuck with NFS or Lustre both of which fail
quite badly with small files.. all this is backed by fibre attached
SAN storage
(they really are temporary and can be easily restored)
James
--
Cosimo
--
The Wellcome Trust Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.
