I have an unusual problem with Solaris 9 servers I am responsible for. Occasionally a server will suffer from very poor performance, characterized by disk throughput ramping up to 100%. I figured there must be a process responsible for hogging the disk, so I first tried shutting down as many processes as I could. The problem is that the servers are distributed around the world, and I need to be careful not to stop so many processes that the server becomes unavailable. This did not yield a solution, so I found some tools to list disk usage by process. I found that when the disk usage is at 100%, there was very little process disk activity, so it was not possible to identify a guilty process. Strangely, one of the partitions affected (according to iostat) is the backup partition. The only solution I have found so far is to do a reboot.