The Question is:
In "Guidelines for OpenVMS Cluster Configurations" appendix D2 you state "For
maximum performance and availability, use two disjoint FDDI LANs, each with
dedicated host adapters and full-duplex FDDI intersite links connected to
GIGAswitch/FDDI bridges at
each site. "
For the last years I have successfully used such configurations to provide high
availability and high performance 3 and 4 site clusters.
One of the components used by our applications is the host based shadowing in
order to provide full disaster redundancy
Now I am being told by Enterrasys that they will stop provide the GigaSwitches
at the end of this year.
1) Is there a prefered path from Compaq concerning the technology used for
designing multpile site OpenVMS Clusters ?
2) What technology and implementations are Compaq using when you qualifying
multiple site OpenVMS Clusters ?
3) Do you have any guidlines for calculating site recovery for multiple site
OpenVMS Clusters ?
Thanks in advance.
The Answer is :
Host-based shadowing and the various other mechanisms and packages that
are used within an OpenVMS Cluster are not particularly specific to the
underlying physical media involved.
The particular network media used and the associated mechanisms are of
rather less interest than are the basic clustering latency and bandwidth
requirements, and meeting the disparate physical media routing and the
required level of redundancy of the physical media. Once the basic
latency and bandwidth requirements are met (eg: 10 megabit Ethernet; the
site-specific requirements will likely require greater bandwidth), you
will be looking at avoiding a single-point failure (such as having the
redundant inter-site network links routed via non-adjacent conduit, and
having multiple redundant network switches or network bridges.)
ATM, Gigabit Ethernet and Fibre Channel are the most common approaches
for remote access in current clustering configurations.
In addition to the clustering information included in the two manuals
in the OpenVMS manual set, disaster-tolerant clustering information is
also available as part of the Disaster Tolerant Cluster Services (DTCS)