 |
» |
|
|
|
OpenVMS
cluster software version 7.3-1 |
Summary of
OpenVMS Cluster Software Enhancments
This section provides an overview of the following new OpenVMS Cluster
Software features that are available in Version 7.3-1.
Fibre Channel
I/O New Features
|
FEATURE
|
BENEFIT
|
|
Multipath Failover to MSCP served path
- Provides automatic and manual failover
from a direct Fibre Channel or SCSI storage path to an MSCP
served path
|
- Provides Cluster-aware
applications the ability to continue running and access necessary
data even if all direct links to the storage have been broken
(automatic failover to the served path)
- Allows a cluster system
manager to manually redirect storage access from a broken direct
path to the cluster MSCP served path for Fibre Channel and SCSI
Storage
- Enhances the already
robust OpenVMS Disaster Tolerance capabilities with a more powerful
failover capability, resulting in access to necessary data regardless
of the type of data access failure in the cluster environment
|
|
Multipath Tapes on Fibre Channel
- Implement multipath for tapes on Fibre Channel
- All possible paths from an Alpha system to a Fibre tape will
be configured and made available
- The exact path will be selectable via a DCL command.
- In the event of a broken connection, automatic failover will
take place
|
- Provides multi-path capability for Fibre
Channel tapes
- Supports the same multipath failover model
for tapes that is supported for FC-attached disks
- Provides for true “High Availability”
model for both Tape and Disk devices in an FC SAN Environment
|
|
Performance enhancements for KGPSA Fibre Channel Adapter
- Multiple device interrupts combined into single interrupt
- Various other driver enhancements for the KGPSA
|
- Improved throughput for frequent cluster
operations due to a more efficient method of handling I/O interrupts
- Less CPU Overhead per I/O on very heavy
I/O loads (up to 25-40% reduction)
- Reduction in overall driver CPU usage by
10-30%
|
Improved
System and Application Performance
|
FEATURE
|
BENEFIT
|
|
Cluster Performance Improvements
- Significant reduction in system overhead
for various components of cluster communications, cluster I/O,
and the VMS lock manager, for both distributed and local locking
|
- Improved throughput for cluster locking,
messages, data transfers
- Reduced system overhead for frequent cluster
operations
- Improved throughput for local locking
- Reduction in nonpaged pool overhead and
image sizes
|
|
