ZFS and storage recordsize mismatch
I'm a bit confused about bandwidth consumption where recordsize mismatch happened between storage controller and zfs.
I have a zpool with recordsize=128KB, this zpool contains only one LUN that resides in STK 6580 storage with segment size = 512KB. I have 2 x 8Gbps FC connections with active standby configuration, so I assume I have 8Gbps bandwidth in total.
Well, so far so good, no performance issues from the application side (Oracle EBS), but how can I calculate the bandwidth consumption of my 8Gbps FC connection ?
When iostat reports 2000 IOPS (total read and write), I assume that it's 2000 x 128KB per second = 256000 KB per second = 1.9 Gbps.
0