We have a SAN connected by two Connectrix at two sites supporting different flavors of platform like Sun, HP, Digital, etc. When we activate a modified zone set, it's normal with a very little disconnection or flick between the switch and the hosts, but it's so small and the hosts (Sun Solaris) only have a warning (message of offline disks and then returned to online) in less than a second. That didn't happen with a Compaq GS160 (OpenVMS 7.2, HBA KGPSA-CA Emulex). When we activate the new zone set, the system crashes. Do you know somebody who is using Connectrix and some OpenVMS system or similar?
Thanks a lot?
In my experience, zone set transitions are instantaneous. You may want to check the switch revision levels to make sure you have the latest version. I have used Brocade and McData switches with OpenVMS and have not seen any errors. I would call in the storage vendor and ask them to review your configuration and zoning procedures. Zone transition time should be in the sub millisecond range and not affect applications or disk access. You should be using dual paths through dual fabrics just in case.
Since you say you are using Connectrix, I assume it is an EMC solution. If you are using CI based clusters with EMC, check the I/O queue depth for the LUNs being accessed in the array. If CI attached, EMC uses a third party emulator for CI based storage traffic. It's best to use pure Fibre for OpenVMS clusters on third party storage. Use CI only on a pure Compaq solution.
Also, if you are updating the zone information over the network and you have network congestion, this may be the reason for the extensive update delay. Check the error log and port error data on the Connectrix for an indication of switch problems. You may also want to run fabric diagnostics to make sure you are not introducing latency from improper cabling. Fibre cables are very susceptible to bends and crimps.
This was first published in June 2001