Chapter 10: Clusters


Overview

OpenVMS has the capability to:

  • Create and manage a cluster of CPUs interconnected with various LAN options.

  • Share resources within the cluster, such as printers and files.

  • Share system management structures, such as user account file (SYSUAF), mail profile, and so on.

  • Support widely dispersed clusters.

  • Support multipathed disk farms.

  • Support system management of all CPUs from a single node.

  • Support diskless nodes.

The OpenVMS Cluster is an OpenVMS feature that permits multiple (up to about 100) computers to be interconnected and to invisibly share resources such as software packages, printers, files, and devices. A cluster also provides a fail-safe operation for the computing facility, because a cluster can be configured to run without a full complement of computers. A cluster might be distributed over several miles and, if the connection between them is severed, the reduced cluster can continue operation (with reduced capacity) in a matter of seconds, as if nothing had happened. Furthermore, once the connection is restored, the cluster melds itself back into a single unit.

OpenVMS out of the box supports clusters. This is not an add-on product and is designed into OpenVMS, although it requires a separate license. Clusters (in the OpenVMS sense) are made up of several computers cooperatively sharing resources. A single copy of the operating system usually resides on one cluster member. All of the members of the cluster are managed and operate under the same security domain (i.e., with a single user authorization database). Thus, the system manager can access any CPU from any other one for management purposes using the SYSMAN command. A user logs in to all CPUs the same way and has the same privileges throughout the cluster. When an application is started, it may execute on a single CPU or, if it is coded correctly, on multiple CPUs. In the latter case the processes must cooperate to access shared resources. A given disk drive can be either made local to one CPU or can be globally read/write accessed by any member of the cluster. In the latter case a clusterwide locking mechanism controls and synchronizes access requests across the cluster.

A cluster is similar to a symmetric multiprocessor (SMP) computer in that most of the cluster's resources can be shared. A cluster is unlike an SMP computer in that each cluster member has its own memory and, optionally, its own disk drive(s). Memory of the individual CPUs cannot be shared, but disks may be. Furthermore, the cluster should be configured so that any member computer can be stopped (e.g., for repairs) and then rebooted into the cluster again without any interruption to the other computers in the cluster. Therefore, critical processes on the stopped computer (e.g., the queue manager) fail-over automatically to a running node.




Getting Started with OpenVMS System Management
Getting Started with OpenVMS System Management (HP Technologies)
ISBN: 1555582818
EAN: 2147483647
Year: 2004
Pages: 130
Authors: David Miller

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net