We found the peak performance with running several seprate threads. Our code set breaks down nicely into 5 separate sections... Time Taken Filesystem Mount Options RAID Sync Write Bias Comments 0:26:20 rw,bg,noxattr,vers=3 Double Parity Latency 0:23:25 rw,bg,noxattr,vers=3 Double Parity Latency 0:21:15 rw,bg,noxattr,vers=3 Double Parity Latency 0:21:26 rw,bg,noxattr,vers=3 Double Parity Latency 0:19:20 rw,bg,noxattr,vers=3 Double Parity Latency 0:20:39 rw,bg,noxattr,vers=3 Double Parity Latency 0:19:58 rw,bg,noxattr,vers=3 Double Parity Latency 0:18:53 rw,bg,noxattr,vers=3 Double Parity Latency 0:20:51 rw,bg,noxattr,vers=3 Double Parity Latency 0:11:52 rw,bg,noxattr,vers=3 Mirrored Latency 0:11:50 rw,bg,noxattr,vers=3 Mirrored Latency 0:12:00 rw,bg,noxattr,vers=3 Mirrored Latency 0:11:30 rw,bg,noxattr,vers=3 Mirrored Latency 0:10:03 rw,bg,noxattr,vers=3 Triple Mirrored Latency 0:00:00 rw,bg,noac,noxattr,vers=3 Double Parity Latency Very Slow - 15M in 10 Minutes - stopped test 0:19:16 rw,bg,noxattr,vers=3 Double Parity Throughput 0:07:30 rw,bg,noxattr,vers=3 Mirrored Latency multi-threaded - 4 threads 0:07:12 rw,bg,noxattr,vers=3 Mirrored Latency multi-threaded - 8 threads
And for comparison running on the same nas using iSCSI (we need NFS becusase the file system is shared)... 0:04:23 iSCSI Double Parity Latency write cache enabled 0:04:15 iSCSI Double Parity Latency write cache disabled -- This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list storage-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/storage-discuss
