Matt Lawrence wrote: >> It is an unfair to compare clustered filesystems and local disk >> filesystems directly. You can try a few optimizations - the io-threads >> is pretty much useless on the client side in 1.4/2.0 branch (since the >> introduction of non blocking sockets). You can remove read-ahead too >> since I understand your IO pattern largely involves random IO, and for >> sequential IO glusterfs can achieve link max speed on Gig/E even >> without read-ahead. After these changes you might also want to try >> with and without write-behind because these performance translators >> are meant to be used with streaming IO. >> >> > > WRF is not the only thing running on the cluster. > > Direct I/O of 1GB takes 2.7 seconds. NFS takes 17 seconds. Gluster > takes 100 seconds. > Pulled those translators out and it reduced the glusterfs time to 77 seconds. A significant improvement, but still far behind NFS. Any other suggestions? -- Matt
