HP OpenVMS Systems

This section briefly describes these features.

SCSI Clusters

Additionally, OpenVMS V6.2/V7.0 provided initial support for SCSI Cluster configurations. However, because SCSI Cluster features have been enhanced in three updates to OpenVMS V6.2 (V6.2-1H1, V6.2-1H2 & V6.2-1H3), this section includes the latest information on SCSI Clusters, based on V6.2-1H3. Note that V6.2-1H3 (order number QA-MT1AG-H8) started shipping in June 1996.

OpenVMS Cluster Client Software

OpenVMS Alpha V6.2/V7.0 introduces a new license type for OpenVMS Cluster software. Called OpenVMS Cluster Client, the new license provides a low-cost cluster client product for Alpha and VAX workstations.

The OpenVMS Cluster Client software provides fully functional OpenVMS Cluster software with two exceptions: (1) Clients may not provide VOTES to the OpenVMS Cluster configuration, and (2) Clients may not MSCP Serve disks, nor TMSCP Serve tapes.

OpenVMS Cluster Client software is available for Alpha and VAX systems. The software uses an LMF license name of VMSCLUSTER-CLIENT. OpenVMS Cluster Client software is included in the Compaq NAS 150 package and is not separately orderable.

OpenVMS Cluster Satellite booting & downline load support without DECnet

OpenVMS Alpha and VAX V6.2/7.0 provide a new feature within the base operating system to perform MOP (maintenance operation protocol) downline loading. In previous releases, this functionality was provided by DECnet Phase IV or DECnet/OSI. DECnet may continue to be used to service MOP downline load operations, or the new feature may be used.

The new MOP downline loading feature is implemented by the LANACP (Local Area Network Ancilliary Control Process). A user interface to the LANACP, LANCP (LAN Control Program), permits a system manager to define and control MOP downline load operations.

The LANACP may be used to service all types of MOP downline load request. This includes OpenVMS Cluster satellite nodes, terminal servers, LAN-resident printers, and X-terminals.

A detailed description of the new LANACP feature can be found in the OpenVMS V6.2/7.0 Release Notes.

DECnet software no longer required for OpenVMS Cluster Software

Prior to OpenVMS V6.2/7.0, DECnet Phase IV or DECnet/OSI software was a prerequisite for OpenVMS Cluster software on VAX (VAXcluster) and Alpha (VMScluster) systems. With the introduction of OpenVMS V6.2/7.0 the requirement for DECnet software is now removed.

DECnet software will be necessary in OpenVMS Cluster environments under the following conditions:

• When applications perform node to node communication using DECnet mailboxes.
• In V6.2, when the MONITOR utility with the CLUSTER keyword or /NODE qualifier is required (in V7.0, Monitor will use TCP/IP or DECnet, whichever is available).

Note that DECnet is not required for the correct operation of SYSMAN in an OpenVMS Cluster configuration. Also, SET HOST operations may be achieved without DECnet by means of the SET HOST/LAT command.

OpenVMS Cluster Availability Manager enhancements (DECamds)

The OpenVMS Cluster Availability Manager includes enhancements to provide additional event filtering and the ability to perform real-time recomputing of OpenVMS Cluster Quorum. Additionally, the Availability Manager console has been ported from VAX, and can now be run on an Alpha system.

For V7.0, DECamds has been enhanced to display many SCS statistics for every node in the cluster, similar to those available in the SHOW CLUSTER utility.

HSJ/HSC and HSD30 series controller failover - enhanced support

In previous releases of OpenVMS Cluster software, dual porting of disks between pairs of HSJ/HSC series controllers was supported when the controllers were attached to a common Star Coupler. With the availability of OpenVMS V6.2/7.0, it is now possible to connect dual ported disks to pairs of HSJ/HSC series controllers that are attached to different Star Couplers. This feature enhances availability because failure of a CI adapter does not make both controllers unreachable.

This feature is available for HSJ/HSC series CI controllers. The feature is also available for HSD30 series DSSI controllers, which may now be configured across different DSSI buses.

Coexistence of OpenVMS VAX and OpenVMS Alpha releases in a VMScluster

OpenVMS Alpha V6.2 and OpenVMS VAX V6.2/7.0 provide two levels of support, called Warranted and Migration, for mixed-version and mixed-architecture VMSclusters.

"Warranted support" means that Compaq has fully qualified the two versions coexisting in a OpenVMS Cluster, and will answer all problems identified by customers using these configurations.

"Migration support" is a superset of the Rolling Upgrade support provided in earlier releases of OpenVMS, and is available for version mixes that are not warranted. Migration support means that Compaq has qualified the versions for use in configurations that are migrating in a staged fashion to a newer version of OpenVMS VAX or to OpenVMS Alpha. Problem reports submitted against these configurations will be answered by Compaq. However, in exceptional cases Compaq may request that you move to a warranted configuration as part of answering the problem.

Migration support will help customers move to warranted OpenVMS Cluster version mixes, with minimal impact on their cluster environments.

Warranted and Migration pairs are supported as follows:

• OpenVMS Alpha V6.2 and OpenVMS VAX V6.2 are Warranted together in a mixed architecture cluster. Any other version pairing, back to OpenVMS VAX V5.5-2 and OpenVMS Alpha V1.5, is supported for Migration.
• OpenVMS Alpha V7.0 and OpenVMS VAX V7.0 are Warranted together in a mixed architecture cluster. Any other version pairing, back to OpenVMS VAX V6.2 and OpenVMS Alpha V6.2, is supported for Migration.

Note that Compaq does not support the use of more than two versions in an OpenVMS Cluster at a time. In many cases, more than two versions will successfully operate together but Compaq cannot commit to resolving problems experienced with such configurations.

Quorum Calculation Changes

In Version 6.2/7.0, the REMOVE_NODE option of Shutdown will correctly calculate a Quorum value for the cluster that reflects the number of Votes remaining after the node has completely shutdown.

This enhancement enables nodes to be shutdown without risk of losing Quorum.

The PCI to CI adapter - CIPCA

The CIPCA, originally supported in V6.2-1H2, is a high-performance CI adapter for PCI-based AlphaServer systems. In V6.2-1H3, it is supported in AlphaServer 2000, 2100, 4100 and 8200/8400 systems. Up to two CIPCAs may be configured in a system.

The CIPCA can coexist on a CI bus with CIXCD and CIBCA-B host adapters. It can also coexist with all HSC/HSJ series controllers except the HSC50.

The CIPCA consists of two modules: a PCI module and an EISA module. When configuring systems with a CIPCA, ensure that one module slot of each type is available.

The PCI to DSSI adapter - KFPSA

The KFPSA, originally supported in OpenVMS V6.2-1H2, is a high-performance DSSI adapter that is supported in all PCI-based AlpaServer systems. Up to four KFPSAs may be configured in a system.

The KFPSA can coexist on a DSSI bus with any DSSI adapter, controller, or device (except the KFMSB host adapter for the DEC 7000/10000 systems).

Fast Path and Fast I/O

These features are available for OpenVMS V7.0 only.

Fast Path is a feature for CIXCD CI-based systems that greatly reduces the compute component of an I/O. With this feature enabled the number of CIXCD port instructions to issue and complete an I/O is reduced by approximately 50%. Additionally, significant improvements have been made to the parallelization of the I/O subsystem. In an SMP CPU with several CIXCD adapters, OpenVMS will balance control and interrupt handling of adapters to different CPUs. This feature greatly increases CPU cache hit rates, and reduces SMP spinlock delays.

Fast I/O refers to three new OpenVMS system services that allow privileged server applications to greatly optimize their I/O. As with Fast Path, the Fast I/O system services reduce the instruction count required to initiate and complete an I/O by approximately 50% (as compared with the $QIO system service).

Miscellaneous Stuff

Several other new features debuted in OpenVMS V6.2: SWXCR device naming change; a new user license type; dumpfile off the system disk; increased volume shadowing limits; ATM support; cluster event services on VAX.

OpenVMS SCSI Clusters - support for multi-host SCSI configurations

OpenVMS Cluster Software for OpenVMS Alpha V6.2-1H3 provides support for multi-host SCSI configurations. The following systems, SCSI adapters, controllers, and disk devices may be configured on multi-host SCSI buses:

• Any PCI-based AlphaServer and AlphaStation system
• Any TurboChannel-based DEC 3000 Alpha system
• SCSI Adapters
  • The embedded NCR 53C810-based SCSI adapter supplied in some of the above systems
  • Optional PCI-based add-on KZPAA adapters
  • Optional PCI-based add-on KZPSA fast wide differential (FWD) adapters
  • Optional TurboChannel-based add-on KZTSA FWD adapters
• Controllers
  • HSZ40 and all its supported disks
• Converters
  • DWZZA
  • DWZZB
• A wide range of SCSI disks may also be directly connected to multi-host SCSI buses (in other words, not behind an HSZ40 controller). Compaq has fully qualified all variants of the RZ26, RZ28, and RZ29 series disks. Moreover, any SCSI-II compliant disk that supports the following optional SCSI-II features will work on a multi-host bus in a SCSI Cluster:
  • Automatic Bad Block Revectoring
  • Tagged Command Queuing
  • Support for multi-initiator I/O requests
    • For more information on the SCSI-II and SCSI Cluster disk requirements, please refer to the OpenVMS New Features Manual

Up to three systems may be connected to a multi-host SCSI bus. A system may be connected to a maximum of six multi-host SCSI buses. Any number of single host SCSI buses may also be configured.