I was under the impression HAMMER was a parallel filesystem. sorry
On Wed, Aug 12, 2009 at 1:04 AM, Matthew Dillon<dil...@apollo.backplane.com> wrote: > > : > :The I/O bottleneck is coming from the disk subsystem and network. I > :was wondering if HAMMER can do parallel filesystem implementation > :similar to GPFS or Lustre. > : > :Also, the reads/writes are random access there is very little > :sequential streaming, but the files are large.Each file is around 30GB > :each > > It can do master->multi_slave replication if that is what you mean. > I don't know how that might compare to GPFS or Lustre. You are going > to have more choices in linux-land then in BSD-land, particularly if > you have a large array of drives. > > If there is an I/O bottleneck from the disks due to random access > seeks the only solution is more spindles. > > If you can stage the data in any way a large SSD (solid state drive) > might help. e.g. one or more 256G SSDs for data staging eliminates > the seek bottlneck and probably also eliminates the need for large > amounts of ram in the machines. > > -Matt > >