From: SMTP%"vms-steering-list-owner@DECUS.Org" 26-OCT-1993 11:38:25.26
To: EVERHART
CC:
Subj: Breton: OpenVMS Futures Primer
Date: Tue, 26 Oct 1993 11:30:36 -0400 (EDT)
From: Betsy Ramsey <RAMSEY@CUA.EDU>
Subject: Breton: OpenVMS Futures Primer
To: VMS-STEERING-LIST@DECUS.ORG
Errors-to: vms-steering-list-owner@DECUS.Org
Warnings-to: vms-steering-list-owner@DECUS.Org
Resent-message-id: <01H4KF1AN4MA8WXMM2@DECUS.Org>
Message-id: <01H4KF0KJJ6C8Y56YG@CU4700.CUA.EDU>
X-VMS-To: IN%"VMS-STEERING-LIST@DECUS.ORG"
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Content-transfer-encoding: 7BIT
Comments: Send vms-steering-list subscribe/unsubscribe requests to
vms-steering-list-request@DECUS.Org
Newsgroups: comp.os.vms,vmsnet.misc
From: breton@star.enet.dec.com (Brian Breton)
Subject: OpenVMS Futures Primer
Message-ID: <1993Oct25.205439.19259@e2big.mko.dec.com>
Organization: Digital Equipment Corporation
Date: Mon, 25 Oct 1993 20:56:57 GMT
OpenVMS Futures Primer
October 1993 Announcement
Digital Equipment Corporation
OpenVMS Engineering
Product Management Office, ZK3-4/T61
110 Spit Brook Road
Nashua, New Hampshire 03062
FAX:+1 603-881-0120 (VMS PM ZK3-4/T61)
Email:OpenVMS_REQ@STAR.ENET.DEC.COM
October Announcement
Copyright Statement
The information in this document is subject to change without notice
and should not be construed as a commitment by Digital Equipment
Corporation. Digital Equipment Corporation assumes no responsibility
for any errors that may appear in this document.
Copyright 1993 Digital Equipment Corporation All Rights Reserved.
The following are trademarks of Digital Equipment Corporation:
OpenVMS DEC FORTRAN
OpenVMS AXP DEC C
VAX ManageWORKS
DECnet DEC C++
OpenVMScluster DEC Pascal
DEC COBOL PATHWORKS for OpenVMS
Preface
This document describes Digital's vision for future feature
enhancements to OpenVMS. Its intent is to explain the concepts of
four major areas.
Part 1 provides a discussion of a new 64-bit file system.
Part 2 provides a discussion of the future plans for OpenVMS 64-bit
support.
Part 3 provides a discussion of enhancing the availability of
OpenVMSclusters.
Part 4 provides a discussion of future plans for OpenVMS system
management software based on a PC management station.
Part 1: The New File System
Overview
Today, OpenVMS provides a robust, reliable, highly-available cluster
file system that meets the stringent data integrity requirements
demanded by customers. The New File System (TNFS) for OpenVMS AXP
will extend these capabilities, especially in the areas of
performance, data management scalability and on-line backup, to meet
the requirements of customers well into the 21st century.
TNFS is a 64-bit clustered physical file system and data protection
utility that uses coherent distributed caches, a revolutionary
log-structured file system technology, and an extended file model.
This technology, based on the latest research work at the University
of California, will go a step beyond contemporary logging file
systems, providing innovative capabilities in the areas of backup,
disaster recovery, clustering, and write performance.
Technology Features
Performance: read performance will equal that of best-in-class file
systems. The sequential structure of TNFS, in combination with its
innovations in caching technology, will break through today's
barriers, yielding leading-edge write performance. These advances in
write performance can be further enhanced when paired with RAID-5.
Continuous Backup: TNFS will perform on-line backup (the ability to
consistently back up data while it is being changed) at an extremely
high speed. The new design makes it possible to leave any or all
previous states of the file system on disk, and then allows recovery
of deleted files, or return to previous versions of files.
Scalability: this new design provides seamless management of up to 10
TB of on-line data. In addition, it will be a foundation for
leading-edge storage management applications by further extending the
addressability of both near and off-line storage.
Disaster Recovery: another benefit of the design is the flexible and
potentially low-cost method of maintaining a remote, consistent copy
of live data.
Cluster Capability: TNFS is designed as a fully-functional cluster
file system, providing transparent cluster-wide access to files. The
availability characteristics of clusters are further extended with the
introduction of a high-performance on-line backup facility, providing
unparalleled scaling and throughput.
Compatibility: existing applications can take advantage of TNFS
without modification because it supports a fully-compatible RMS file
system API and an equivalent backup interface. Further, technologies
like clustering and volume shadowing utilize TNFS features
transparently.
Heterogeneous Client Support: TNFS is designed around an extended file
model that is a superset of the file features used by the industry's
leading operating systems. TNFS enhances the capabilities of
distributed file systems servers, allowing all popular clients access
to the same file. For example, a single file accessed by
heterogeneous clients such as UNIX, Windows NT, OpenVMS, and
Macintosh, will appear to each as a native file.
Summary
OpenVMS customers can be assured that OpenVMS will continue to support
them by reliably managing their data into the 21st century. The new
file system supports terabytes of on-line data with high performance
and high availability while enabling rapid recovery from data loss or
disasters.
Part 2: OpenVMS AXP 64-bit Plans
Overview
Digital is moving toward a full 64-bit virtual addressing
implementation on OpenVMS. 64-bits for OpenVMS means much greater
performance and data capacity for applications that choose to use
64-bit addressing. Examples include molecular modeling, imaging, and
real-time voice analysis.
At the same time, OpenVMS engineering is committed to continuing to
provide a 32-bit environment for applications that have no need for
64-bit virtual addressing, therefore, 32-bit applications will NOT
need changes in the future. This preserves investment in today's
applications as customers move them to Alpha AXP systems.
Several of Digital's compilers on OpenVMS AXP already take advantage
of 64-bit technology through an efficient internal representation of
18 digits of packed decimal data. For example, the COBOL compiler for
Alpha AXP systems will transparently convert packed decimal data upon
input/output to/from this new representation. Packed decimal
calculations are then done very quickly with the new format. This
means that COBOL applications will have far better performance on
Alpha AXP systems.
Below is a table showing the data declarations needed to access 64-bit
integer and floating point for each of Digital's compilers.
64-bit 64-bit
Language Integer Floating Point
DEC FORTRAN V6. INTEGER*8 REAL*8
DEC C V1.3 __int64 double
DEC C++ V1.3 __int64 double
DEC Pascal V5.0 INTEGER64 DOUBLE
DEC COBOL V1.1 PIC 9(18) USAGE COMP USAGE COMP-2
OpenVMS AXP memory management code has been implemented without the
30-bit physical memory restriction. It is now possible to put
gigabytes of memory onto an OpenVMS AXP system.
In addition, OpenVMS AXP supports up to one terabyte files and
databases today, easing restrictions on maximum file size. While this
is not technically a "64-bit" implementation, it is consistent with
the scaling of OpenVMS toward greater performance and data capacity.
Alpha AXP systems bring the power of RISC computing to OpenVMS
applications. As the 64-bit capabilities unfold, OpenVMS AXP will set
the stage for a whole new generation of applications.
Part 3: Enhancing the Availability of OpenVMSclusters
Overview
VAXclusters have provided an industry benchmark in availability,
scalability, ease of management, and configuration flexibility.
Today, VMSclusters can be configured into highly available systems
with no single points of failure, with redundant paths to I/O devices,
data redundancy using volume shadowing, and application recovery with
RMS Journaling. With this announcement, OpenVMS- clusters, consisting
of OpenVMS VAX systems and OpenVMS AXP systems, share the same
attributes and features. OpenVMS AXP systems now support Volume
Shadowing to ensure access to data even in the event of disk failure
and RMS Journaling for application recovery. OpenVMS AXP systems now
also support the FDDI as a cluster interconnect.
OpenVMSclusters provide an ideal platform for building highly
available, scalable servers unmatched by any other platform in the
industry. However, providing a highly available system is only part
of creating a high availability server. With PATHWORKS V5 for
OpenVMS, the full range of OpenVMScluster scalability and availability
features can now be incorporated into all client/server applications
running on popular PC clients: DOS, Windows, Windows NT and OS/2. If
a network connection to a cluster member goes down, a new connection
is automatically made to another cluster member. When a PATHWORKS
session is created, client connects are targeted to the least loaded
cluster member. PATHWORKS cluster support is transport-independent.
You can use any of the supported PATHWORKS for OpenVMS transports:
DECnet, TCP/IP, and NETBEUI.
One of the most overlooked aspects of increasing system availability
is eliminating outages caused by human error. Studies have shown that
between 25-35% of all system outages are caused by human error.
ManageWORKS, available with PATHWORKS for OpenVMS V5, manages
multivendor NOS environments including PATHWORKS servers and clients,
NetWare servers, and LAN Manager servers. ManageWORKS provides a
graphical user interface and a hierarchical management view to enable
users to manage in a meaningful way. In the near future, the OpenVMS
operating system will support system management software based on
ManageWORKS, allowing for integrated and consistent management of both
OpenVMScluster systems and PC LANs.
Part 4: The OpenVMS Management Station Software
Overview
The OpenVMS management station software currently under development
will enable management of one or more OpenVMScluster systems from a
single Windows-based PC. This is the model for OpenVMS system
management of the future.
The project goals are:
- to simplify and reduce the effort required to manage one or more
OpenVMScluster systems
- to provide the ability to manage one or more OpenVMScluster systems
as a single management and user domain, out-of-the-box
- to scale the effort required to manage an OpenVMS site linearly and
predictably relative to increasing size and complexity
- to eventually be able to manage systems remotely
The management station software is based on the PATHWORKS V5.0
ManageWORKS framework. This graphical user interface provides the
infrastructure for implementation of system management functions such
as user account management or printer management. The ManageWORKS
framework presents a hierarchical management view and provides the
ability to manage multiple OpenVMS systems from a single source. It
enables users to view, organize, and manage objects in a way that is
meaningful to them. The interface will provide an integrated,
consistent look-and-feel for all OpenVMS management tools as opposed
to the varied syntax of the DCL interface for each utility today. The
target for the PC management station software is OpenVMS multi-cluster
sites and in particular, environments consisting of PC LANs with
OpenVMS hosts. While the PC management station software is intended
to manage only the OpenVMS environment (it is not designed to be an
enterprise director), the ManageWORKS interface brings both LAN and
OpenVMS management together under one framework and provides the
opportunity for integration of applications and services.
The first version of the management station software will provide the
capability to manage user accounts and the infrastructure under which
future management applications will be deployed.
---------------------------------------
Brian Breton, Breton@STAR.enet.dec.com
OpenVMS Product Management
Digital Equipment Corporation
110 Spit Brook Road
Nashua, NH 03062
USA
---------------------------------------