|
HP OpenVMS Systemsask the wizard |
|
The Question is: I have a existing CI cluster running 2 x ES40s on V7.1-2, with redundant HSJ50s on V54 in 2 x SW800 disk cabinets. I would like to split the above up into 2 sites, with each site consisting of 1 x es40, 1 x star coupler and 1 x SW800 (with its own redundant HSj50s). The media between the 2 sites will be via fibre link. Without having to junk any of the existing hardware but also able to mirror disks across the fibre link (fddi) between the 2 sites, what sort of VMScluster configuration are we looking at? The nearest I got so far is the multi-mode VMScluster that was mentioned in the Cluster Manual. Is this the only way to go? What about the latest VMS 7.2-1 - we are willing to explore upgrading to this VMS version or Galaxy. Anything in the pipeline? Rgds Mike The Answer is : The requirements for OpenVMS Cluster communications are quite simple on the surface, each OpenVMS Cluster member must have a direct communications path with every other OpenVMS Cluster member, and the minimum acceptable latency and bandwidth for the path must match that (or better that) provided by a traditional Ethernet (1 megabyte per second) network connection. This includes the ability to use an FDDI or ATM or T3 link as a point-to-point LAN connection between "lobes" of the OpenVMS Cluster. (Connections such as a 56 kilobit dial-up or a 1.54 megabit T1 point-to-point connection are insufficient for OpenVMS Cluster operations.) This requirement for total connectivity can be met via point-to-point connections, or via via a shared (local, bridged, or switched) LAN environment, or via a combination of multiple interconnect technologies. Host-based volume shadowing can span "lobes" of an OpenVMS Cluster. Consideration must be made to the requirement for achieving and maintaining quorum within this configuration -- please see the OpenVMS FAQ for a discussion of VOTES and EXPECTED_VOTES. Additional requirements around the correct operation of the cluster during a (potential) failure of the local (or remote) lobe must also be considered -- this centrally involves the particular voting configuration and operator procedures, and can include the configuration of additional (voting) nodes. OpenVMS Galaxy is an operating environment with multiple instances of OpenVMS coexisting within one physical (SMP) system enclosure. OpenVMS Galaxy configurations by definition do not span sites, though OpenVMS Cluster configurations comprised of multiple standalone and/or multiple OpenVMS Galaxy systems can certainly span sites. As it appears you are unfamiliar with OpenVMS Cluster configurations and operations, please consider contacting Compaq presales technical support or your Compaq reseller for assistance -- there can be far more to this discussion and to your specific environment and to your specific local requirements than can be included here...
|