We are experiencing a problem with performance on our storage pool. We created a 10TB lun but we are experiencing slow write speeds. We have 3 PAC SAN units with 48 1TB drives a piece. Due to a limitation on how many luns we can have and no JBOD support, we created 36 raid 0 stripes consisting of 4 drives a piece. We then shared those drives to our solaris box and created a zfs pool consisting of 6 raidz2 sets of 6 luns a piece from the PAC storage. I am attaching a picture of our environment, It just shows 1 bay but we have 3 configured. http://www.pacdata.com/index.php?page=971155.txt&id=24 We have 2 FC interfaces connected, 1 from each controller. I checked and MPxIO is set to round-robin on the load balance. Should I change that to a different variable? The stripe size is currently set to 1024KB and the chunk size is 256 KB. What would I need to change, either on zfs or the storage to bring the stripe size back in line? I have live data on this box, so moving the data to other storage would be hard. -- This message posted from opensolaris.org -------------- next part -------------- A non-text attachment was scrubbed... Name: Solaris Storage.jpg Type: image/jpeg Size: 68742 bytes Desc: not available URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20100317/f8f380b7/attachment.jpg>