12h disk array

[Mark Mitchell]

I have a 12h disk array. Out of the box it is set up with 2 controlers. 1 is the primary and the other is its backup. The 1st sees activity and the second lies dormant until a failure of the first. If I go in through lvm and set up half the vg's using the second cards addresses to force the io though there, will it work like a mod array will or will it cause problems?

[Ron Cornwell · Accepted solution]

I have a similar setup. R/390 w/ 2 FWD Controllers, each controller hooked up to a 96MB Controller in the 12H. If you do an ioscan -fnC disk you will see that each LUN is seen twice (once for each controller). What I did was create alternate links to each of my LUNs, then removed the primary link from half of the LUNs forcing them to use the other controller, then establishing the old primary link as an alternate link for those LUNs. I can now look at statistics from measureware and see that the load is closely balanced between the two controllers. HTH

[Mark Mitchell]

I give this a try
thank you

[Bill McNAMARA_1]

Hi Mark, You don't really want an active passive scenario. You'd prefer to do as you say, some lvm primaries through CX and some through CY. Then with two PVs, on through X one through Y, create a VG. Then stripe your LV across these two PVs. Otherwise accessing an LV will still be limited to access throu X exclusively or Y exclusively.
If you look in the mass storage - disk array forum you'll find lots on hte autoraid configuration.

Important for autoraid though is to use some of the disk space to keep data in R10, You don't want to use all of the space possible or.. the data will all be in R5 effectively. Use active hot spare so that space is reserved for R10. This is where you'll see a large perf benefit
Later,
Bill