2016-11-21 05:43 AM
IHAC who has 2 NetApp clusters (primary and DR site). There are mirrored zpools on Solaris 11.3 (OracleDB production), each mirror has one LUN from primary site and one from DR site. Looks like this:
scan: none requested
NAME STATE READ WRITE CKSUM
testpool ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
c0t600A09804D53346E7A3F486F4C315077d0 ONLINE 0 0 0
c0t600A09804D53346E7A3F486F4C315078d0 ONLINE 0 0 0
well, now the question: There are SSD disks (aggregate) on primary site and SAS on DR site. Is there any option to "force" the pool to use primary site (SSD) LUN for read or anyhow set higher LUN priority? It's clear to me that writes IOPS go to both LUNs.
Solved! SEE THE SOLUTION
2016-11-22 12:54 AM
Actually the pool layout (mirror of 2 LUNs, each on one site) stayed the same as before. NetApp controller already uses SSD cache (flashpool) on DR site, so we did not decide to add additional cache on ZFS...