On Dec 29, 2009, at 11:26 AM, Brad wrote:
@relling "For small, random read IOPS the performance of a single, top-level vdev is performance = performance of a disk * (N / (N - P)) 133 * 12/(12-1)= 133 * 12/11where, N = number of disks in the vdev P = number of parity devices in the vdev" performance of a disk => Is this a rough estimate of the disk's IOP? "For example, using 5 disks @ 100 IOPS we get something like: 2-disk mirror: 200 IOPS 4+1 raidz: 125 IOPS 3+2 raidz2: 167 IOPS 2+3 raidz3: 250 IOPS" So if the rated iops on our disks is @133 iops 133 * 12/(12-1) = 145 11+1 raidz: 145 IOPS????? If that's the rate for a 11+1 raidz vdev, then why is iostat showing about 700 combined IOPS (reads/writes) per disk?
Because the model is for small, random read IOPS over the full size of the disk. What you are seeing is caching and seek optimization at work (a good thing). But, AFAIK, there are no decent performance models which take caching into account. In most cases, storage is sized based on empirical studies. -- richard
r/s w/s Mr/s Mw/s wait actv wsvc_t asvc_t %w %b device 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 c0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 c0t0d0 1402.2 7805.3 2.7 36.2 0.2 54.9 0.0 6.0 0 940 c1 10.8 1.0 0.1 0.0 0.0 0.1 0.0 7.0 0 7 c1t0d0 117.1 640.7 0.2 1.8 0.0 4.5 0.0 5.9 1 76 c1t1d0 116.9 638.2 0.2 1.7 0.0 4.6 0.0 6.1 1 78 c1t2d0 116.4 639.1 0.2 1.8 0.0 4.6 0.0 6.0 1 78 c1t3d0 116.6 638.1 0.2 1.7 0.0 4.6 0.0 6.1 1 77 c1t4d0 113.2 638.0 0.2 1.8 0.0 4.6 0.0 6.1 1 77 c1t5d0 116.6 635.3 0.2 1.7 0.0 4.5 0.0 6.0 1 76 c1t6d0 116.2 637.8 0.2 1.8 0.0 4.7 0.0 6.2 1 79 c1t7d0 115.3 636.7 0.2 1.8 0.0 4.4 0.0 5.8 1 77 c1t8d0 115.4 637.8 0.2 1.8 0.0 4.5 0.0 5.9 1 77 c1t9d0 114.8 635.0 0.2 1.8 0.0 4.3 0.0 5.7 1 76 c1t10d0 114.9 639.9 0.2 1.8 0.0 4.7 0.0 6.2 1 78 c1t11d0 115.1 638.7 0.2 1.8 0.0 4.4 0.0 5.9 1 77 c1t12d0 1.6 140.0 0.0 15.1 0.0 0.6 0.0 4.4 0 8 c1t13d0 1.3 9.1 0.0 0.1 0.0 0.0 0.0 1.0 0 0 c1t14d0 -- This message posted from opensolaris.org _______________________________________________ zfs-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
_______________________________________________ zfs-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
