| Here is the article I discussed in reply .6. See how this compares to
the business week article.
Excerpts from a report written by Marc Schulman at Salomon Brothers:
==========================================================================
IBM AND DEC:
THE BATTLE FOR SUPREMACY IN THE COMPUTER INDUSTRY
Summary and Conclusions
IBM and Digital Equipment Corp. (DEC) are adversaries with common
interests. They are adversaries because DEC seeks to supplant IBM as
the primary supplier of computer systems to large organizations. They
have common interests because their dominant positions are threatened
by the emergence of a Sun Microsystems - and AT&T - led alliance of
vendors whose goal is to undermine their market share by popularizing
computer systems based on the Unix operating system. Because Unix has
emerged as the most popular operating system for workstations and
workstation demand is growing at a faster rate than minicomputer and
mainframe demand, the Unix threat and the "downsizing" threat are two
names for the same phenomenon.
In this report, we describe and analyze the strategies that IBM and
DEC will pursue against each other and against their shared threat.
Our primary conclusions are, first, that DEC will continue to gain
market share at IBM's expense at least until such time as IBM
jettisons its System/370 architecture - an event that is unlikely to
occur before the mid-1990s, and, second, that the ability of the Unix
alliance to penetrate DEC's and IBM's customer bases will begin to
diminish within 6-12 months.
IBM Versus DEC
Both IBM and DEC seek to become the dominant vendor of company-wide
integrated computer networks. That this is both companies' objective
is beyond dispute. What is open to dispute is the strategies each
will follow in pursuit of this goal. Because DEC's success in
penetrating IBM accounts has resulted in the proliferation of its
minicomputers and departmental networks, it is widely believed that
the company will continue to pursue a minicomputer-based strategy.
IBM has responded to DEC's inroads by endorsing the concept of
departmental computing and introducing new minicomputer systems - the
9370 and, soon, the Silverlake - thus, many have concluded that the
minicomputer will be the primary battleground between the two
companies. If this conclusion is reached, two issues automatically
arise: First, will IBM be able to stem the erosion of its mid-range
market share; and, second, to what degree will IBM and DEC be hurt by
the displacement of minicomputer - base departmental system by local
area network (LAN)- based departmental systems.
We believe that the minicomputer plays a relatively minor role in
IBM's and DEC's strategies; consequently, investors who are focusing
on the future trends in their minicomputer market shares and the
substitution of LAN-based departmental systems for minicomputer-based
departmental systems are concerned with the wrong issues. In our
view, the battle between the two companies will be fought at both the
high end and at the low end;
o IBM's strategy is to reduce - not enhance - the minicomputer's role,
by encouraging the migration of departmental data bases to mainframe
computers and the migration of departmental program execution to
workstations. Support for this contention includes the recent
announcement of Enterprise Systems Architecture (ESA)/370 and IBM's
intention to introduce very powerful RT PC workstations. By
facilitating the centralization of data bases, ESA/370 represents and
attempt to boost mainframe demand by redefining the role of the
mainframe to that of a corporate data repository from that of an
application program executor. Thus, the mainframe assumes a role that
cannot be occupied by smaller processors. This functional
differentiation should facilitate the maintenance of premium prices
for mainframes. In combination with IBM's Systems Application
Architecture(SAA), ESA/370 provides the architectural foundation for
cooperative processing between mainframes, personal computers (PCs)
and workstations. As its customers implement applications involving
cooperative processing, IBM's goal of having its low-end processors
stimulate demand for its high-end processors should finally begin to
be realized.
o DEC's strategy is to expand its product line through the
introduction of significantly more powerful uniprocessors and
symmetric multiprocessors (SMPs). This strategy began to unfold with
the introduction of the VAX 8800 Series in March. In late 1988 or
early 1989, we expect DEC to announce uniprocessor and multiprocessor
versions of its Argonaut processor. We believe that the performance
of the Argonaut will be more than double that of the VAX 8700 - the
uniprocessor upon which the VAX 8800 Series is based. As evidenced by
the fact that the performance of DEC's SMPs is maximized in mixed
application environments, they are intended to meet all computing
requirements - ranging from office automation to transaction
processing to real-time processing - for up to 500 people. Like IBM,
DEC views network-connected PCs and workstations as sources of demand
for the services of high-end processors. This is evidenced by the
announcement of the Network Application Support (NAS) program in
January and the forthcoming formal introduction of DECwindows.
DECwindows provides the foundation for cooperative processing between
the VAX, IBM and IBM compatible PCs and the Apple Macintosh.
The key differences between IBM's and DEC's strategies are the roles
assigned to high-end processors and the range of low-end processors
that are supported. In IBM's strategy, data is centralized at the
high end, program execution is distributed at the low end, and
cooperative processing involves the transparent movement of data
between high-end and low-end IBM processors. In DEC's strategy, both
data and application program execution are distributed, and
cooperative processing involves program segmentation and data movement
among DEC and non-DEC processors. WE believe that DEC's strategy will
enable it to continue to take market share from IBM. Our reasons are
as follows:
o While IBM's installed base in major corporations is concentrated in
financial and administrative functions, DEC's is broadly distributed.
Years ago, DEC achieved a significant penetration of manufacturing and
development divisions. Recently, the company broadened its
penetration to include such commercial application as office
automation.
o Company-wide networks are being created from the bottom up, not the
top down. Departmental networks are being combined into divisional
networks, and divisional networks are being combined into company
networks. We believe that the greater breadth of DEC's penetration
means that it is better positioned to act as an integrator of separate
networks. We also believe that department and division managers will
resist attempts at data base centralization.
o Apple Computer's success in penetrating the business market means
that many corporations now have both IBM and Apple PCs. Most
corporations will want both types of PCs to be connected to the same
network. We doubt that IBM will support Apple PCs, because it would
hurt demand for IBM PCs. Because it does not compete in the PC
market, DEC is not similarly constrained.
o Even in the unlikely event that the price/performance and
functionality of IBM's mainframes were enhanced to match those of
DEC's SMPS, DEC would retain a significant advantage-much lower
support costs. Because of the complexities of the System/370
architecture, each mainframe installation requires the full-time
presence of professional programmers. DEC's VAXes do not impose such
a requirement. As networks grow in size, he support cost differential
widens.
o By the mid-1990s, IBM may be in a position to replace the System/370
with more user-friendly architectures. When fully implemented, SAA
will break the ties between machine architectures and application
programs. Hence, SAS will provide a means by which IBM can change
architectures without obsoleting its customers' software investments.
[...]
IBM and DEC VERSUS UNIX
The Unix operating system was created by AT&T in 1970. Its existence
was barely visible until the early 1980s, when a legion of start-up
companies opted to pay a license fee to AT&T in lieu of developing
proprietary operating systems. By utilizing Unix, they reduced their
product time to market, and the amount of venture capital required to
fund product development. Although Unix is a multiuser operating
system, start-ups that utilized Unix for multiuser systems were
largely unsuccessful. Instead, it was start-ups that employed Unix in
single-user workstations that met with success. Thus, the popularity
of workstations is responsible for the popularity of Unix.
Few, if any, people claim that Unix is superior to the proprietary
operating systems available from IBM, DEC and others. In fact, Unix
suffers from a number of shortcomings. Despite these shortcomings,
Unix has become a force to be reckoned with, largely because of a
successful proselytizing effort by Sun Microsystems. Sun has
positioned Unix as the means by which computer users can escape their
enslavement to proprietary operating systems. One component of its
message has been that, because Unix and only Unix is available from
many vendors, it follows that the only way to create multivendor
networks is to have all of the systems on the network run Unix.
To date, Sun has been right. Although IBM and DEC offer Unix-based
systems, neither has integrated Unix into its computing environment.
Hence, neither has provided a reason for their customers to purchase
Unix-based systems from themselves, rather than from other vendors.
As a result, Sun and other Unix vendors have rapidly penetrated IBM's
and DEC's customer bases. By this time next year, both IBM and DEC
will have integrated Unix into their respective computing
environments. When this occurs, other vendors are likely to find it
far more difficult to sell to their customer bases.
o DEC's response to the Unix threat is more far reaching. As part of
its DECwindows program, which we expect to be formally announced in
the late summer, DEC will support interoperability between VAXes
running its proprietary VMS operating system and VAXes running Ultrix
(DEC's version of Unix). When this occurs, VMS-and Ultrix-based VAXes
will be able to share files and engage in cooperative processing.
Because other vendors' Unix-based systems will not be supported by
DECwindows, DEC will provide its customers with an incentive to buy
their Unix products from DEC. The exclusion from DECwindows of other
vendors' Unix-based systems does not mean that DEC's VMS-based systems
will not be able to participate in multivendor networks. DECwindows
is a proprietary extension of X-Windows, a set of industry-standard
protocols relating to graphical interfaces. In the next generation of
graphics products, VMS's current graphics protocols will be replaced
by the X-Windows protocols. When this done, VMS-based VAXes will be
able to participate in multivendor networks. What will be missing is
the added functionality provided by DECwindows.
o In its recent strategic endorsement of Unix, IBM announced that the
AIX operating system will be supported on 3090s, 4381s and 9370s in
March 1989. AIX is the same version of Unix that runs on IBM's RT PC
workstations and the PS/2 Model 80. In July, IBM will release its
Common Programming Interface(CPI)for AIX. Applications conforming to
CPI will be able to run on System/370s, RT PCs and the Personal
System/2 (PS/2) Model 80. IBM is erecting a barrier to the
penetration of its customer base, because only AIX-based workstations
and PCs will have applications commonality with its
System/370-architected minicomputers and mainframes. Unlike DEC, IBM
has not indicated as intention to provide interoperability between
systems running its version of Unix and systems running its
proprietary operating systems. Although IBM has endorsed X-Windows,
it has not indicated an intention to alter the graphics protocols of
its proprietary operating systems to support it. Hence, we believe
that the only IBM processors that will support X-Windows will be those
running AIX.
Thus far, we have focused on the measures that IBM and DEC will take
to defend their customer bases from being penetrated further by other
vendors' products, and, in DEC's case, the measures that will provide
for interoperability between its VMS-based systems and Unix-based
systems. In addition, both companies will be introducing more
powerful workstations.
o IBM has publicly stated that it will introduce an RT PC in 1989 that
will be more than four times as powerful than the current RT. Based
upon benchmarks recently released by an independent performance
testing organization (Workstation Laboratories located in Dallas,
Texas), we estimate that the RT PC to be introduced next year will
have twice the current performance of Sun's most powerful workstation
- the 4/260/ This summer, we expect DEC to introduce a workstation
using two CVAX chips to produce processing power equivalent to that of
a Sun 4/260. We expect this product - code-named Firefox - to be
priced at $60,000-70,000. Next year, we anticipate that DEC will
introduce its first workstation employing a reduced instruction set
(RISC) architecture. We believe that the workstation will be capable
of executing 15 million instructions per second (MIPS).
o Recently, DEC's management stated that an SMP version of the Ultrix
operating system will begin testing this fall. If SMP capabilities
are implemented in Ultrix as they are implemented in VMS - entirely in
software - then the SMP version of Ultrix should be able to run the
RISC workstation. Hence, we believe that there is a good chance that
DEC will introduce a four-processor version of the workstation
sometime in 1989. The performance of DEC's product would be similar
to that of Apollo Computer's recently announced Series 1000,
deliveries of which are scheduled to begin in the fall. We are not
aware of any IBM plans to introduce SMP versions of RT PC.
|