[Lustre-discuss] is lustre for us?

[Andreas Dilger]

On Jul 05, 2008  13:36 -0400, Mag Gam wrote:
> At my university we have a big storage problem. Currently we have 25 Linux
> servers each equiped with 3 TB of data. Totaling to 75TB of storage. We use
> each server as a NFS server and have clients mount up the filesystem. They
> are all connected by a GigE network.
> 
> The problem we are facing is performance and scalibility. When a users
> access data, many times they access the same server. This is becoming a huge
> performance problem. The files sizes range from 8GB to 70GB.
> 
> 
> Would Lustre solve our performance and management problem? How would lustre
> scale? I have been reading the quickstart guide, is it easy to implement?

As a general rule, Lustre performs very well with large files and linear
IO patterns, which seems to match your usage.  You can get very good
scaling (nearly linear with each OST) and a single client+server can
saturate a GigE network (110MB/s).

Configuring Lustre 1.6 isn't much different than NFS.  Format each server,
mount them, then mount the clients.

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.