[Bioclusters] Re: blast and nfs

[Chris Dwan (CCGB)]

> given the (ever increasing) sizes of databases bioinformatics software
> is run against - what size of local space would you recommend? 

(Like Joe says) Right now the "knee" of the price/performance curve on IDE
disks is between 60 and 100GB.  Software RAID-0 across two of those will
give you plenty of throughput and space to work with.

Keep in mind that we're not physics, we don't have to deal (yet) with
terabyte datasets and petabyte experiments.  40GB is more than enough
to cache all of the publically available sequence data uncompressed.
120GB (two of the above devices) provides plenty of breathing room.

-Chris Dwan
 University of Minnesota