Originally posted by: danr
We implemented HUR between two USPVs using an HDS developed design for our stated requirements. As soon as the load grew the cache write pending percentage shot up to 70% causing dramatic slowdown of our business applications. We thought the pair(s) causing the write pending rate to soar would split as that was our previous experience with TrueCopy when the load exceeded what the box could handle at one time but a split never occurred in HUR.
To get our applications running at a reasonable rate again we had to figure out ourselves which pairs were creating the high activity and manually split those pairs for periods of time. HDS first told us to create a CLPR to resolve the issue, then add more array groups dedicated to journaling, and then to configure unused ports as RCU targets and initiators on each side.
Has anyone else experienced difficulty with HUR and have you resolved the issues? What were your problems and solutions?