I''m hoping someone has already come across this problem and solved it. I''m using xVM to create 2 NFS fileservers. The Dom0 has a zpool with 8 zvols in it. 4 of the zvols are used for 4 DomU. 2 of the DomUs are fileservers, and each fileserver attaches 2 more of the zvols for the user filespace. The user filespace is then exported via NFS. The zvols attached as the user filespace are formatted UFS. For the first 25 or 30 nfs clients to mount the exported filespace performance is OK, however it then drops off dramatically - 90 seconds to do an "ls -l" I''ve done no tuning, everything is stock. Would using zfs for the user filespace rather than UFS help? Any other ideas for a) determining where the problem is b) improving the throughput. Thanks john Landamore Dept Computer Science University of Leicester UK