This problem took us and Sun a very long time to sort. It turned out that the lower revision of Seagate drive firmware has a bug which knocks off the 1GB cache of the T3. The solution was to replace all revision 04 drives with revision 05. Throughput went from 66MB per sec to 98 MB per sec when all the high rev. drives were used. Solution: Dont mix Seagate: ST336605Fdrives with ST336704F in the same T3 tray. >> Config: 4 x T3 trays configured in partner pairs 4 x T3 Lun's configured and showing in Solaris 8 1 x VxVM Volume striped down 4 columns and using all trays 2 x pairs directly connected to HBA on Fire4810 There are no errors on the T3's or in the syslog. Our problem is read performance on the volume. It is used for business objects reporting. 3 of the trays have extactly the same disks: 36GB Seagates : ST336605F and the other tray has 36GB Seagates: ST336704F. The follow table shows the performance comparison: T3 Tray KB/s Service (ms) T30_0 20447.4 50.2 T30_1 34317.9 30.4 T31_0 48971.5 20.9 T31_1 47174.8 21.7 This is driving us all round the bend. Can anyone shed some light on why this is happening. We need to ensure that IO is balanced across all four trays as it should be with a stripe. Thanks very much in advance. Kev Smith << _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagersReceived on Wed Sep 24 08:38:58 2003
This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:43:23 EST