HP OpenVMS Systems

ask the wizard
Content starts here

OpenVMS Cluster, Volume Shadowing?

» close window

The Question is:

I want to have on on-line contingency between these 2 servers. The alternative
 should be able to fall over to another server with the most updated
 information processed.
We know they can both cluster together but are not sure if the volume shadowing
 product is able to accomplish what we expect. I was told that both servers
 require to have exact disk types/configurations.
Is this correct ? If so that means I can't use our old 2100 for a fall over
 contingency without losing a complete day of transactions.

The Answer is :

  It appears you are mixing the typical fully-live operations expected
  of an OpenVMS Cluster configuration with the a primitive failover-based
  With OpenVMS and OpenVMS Cluster configurations, all servers are
  active and all are operating in parallel, with a configuration that
  is normally designed to provide graceful degradation on failures.
  (This means you get some use out of all of your servers, and you
  know when you are operating in a degraded mode, and you know that
  all of your servers are functioning -- an attempt to failover to
  a server that is discovered non-functional is a situation to be
  avoided, obviously.)
  Host-based volume shadowing provides remote shadowing (mirroring) and
  full write-sharing within an OpenVMS Cluster environment.  This avoids
  the degradation problems that can arise when a RAID controller becomes
  unavailable, because the host-based shadowing product permits the data
  to be shadowed (mirrored) across servers, and across geographic areas.
  Data Replication Manager (DRM) is the closest hardware analog to the
  capabilities of host-based volume shadowing, and there are differences.
  DRM typically supports wider geographic distributions than does the
  OpenVMS Cluster configuration, but shared and parallel access to the
  various mirrored volumes is not available; all I/O access must go to
  one volume.
  As for failover, the hardware configuration need only be identical if
  the failover processing or if the local software requires it -- in the
  more primitive failover-based schemes, similar or even duplicate
  configurations can be necessary.  In the case of a typical OpenVMS
  Cluster configuration, each server is already configured and running,
  so there is no particular need to have similar hardware configurations.
  You do, of course, need to have sufficient processing and storage
  resources available for any operations during the degraded (failure)
  mode environment.
  You will want to review the available materials on the OpenVMS Cluster
  software -- there is extensive documentation in the OpenVMS manual
  set -- and you will want to review the available materials on clusters
  and at the related Disaster-Tolerant Cluster Services (DTCS) details
  available at:
  Compaq can offer services to help you configure your cluster for
  disaster-tolerance, with experience in ensuring the necessary power
  sources are available, proper networking hardware and communications
  cable routing, lighting, graceful failover policies and procedures,
  and related -- designing and setting up an OpenVMS cluster can be
  quite easy; properly designing and implementing a disaster-tolerant
  configuration is seldom a trivial undertaking.  The DTCS Services
  can help.

answer written or last revised on ( 29-APR-2002 )

» close window