Behind the News
Analysis of Industry Events
Tivoli Turns Blue
All pieces of printed marketing literature for IBM's SystemView products
invite potential customers, "When you think systems management, think
IBM." Now that IBM is in the process of acquiring distributed systems
management framework and tools vendor Tivoli Systems of Austin, TX, managers
of distributed open systems are more likely to think of IBM when considering
the purchase of such products. Although SystemView for AIX (the IBM version
of Unix) has incorporated some network management framework technology licensed
from Hewlett-Packard, the current SystemView product line is a disparate
collection of platform-specific systems management tools for MVS mainframes,
AS/400s, PCs running the OS/2 operating system and RS/6000s running AIX.
"SystemView lacked any consistent agent technology, and its common
launch pad does not provide a meaningful level of integration," says
Chet Geschickter, vice president of Hurwitz Consulting Group of Newton,
MA, in a published report on IBM's acquisition of Tivoli.
By IBM's own admission, SystemView lacked the breadth that competing systems
management solutions, such as CA-UniCenter from Computer Associates and
HP OpenView, can provide. "IBM SystemView already has the tools. What
we lacked was an object-oriented framework to link them," says Alfred
Zollar, vice president for SystemView at IBM's systems management group
in Research Triangle Park, NC. "There is synergy in this merger, because
we and Tivoli share a vision of how system management applications should
be delivered." (IBM consistently refers to its cash tender offer of
$743 million for Tivoli as a merger.)
Predictably, Tivoli echoes this view. "IBM has not changed its SystemView
architecture. Rather it has acquired an implementation of it," says
Martin Neath, vice president and general manager for the Tivoli Management
Environment (TME) core products.
For Tivoli, becoming part of IBM is in some ways a return to its roots.
The company was started in 1989 by two former IBM employees who developed
the idea of an object-oriented systems management framework while working
at IBM labs in Austin. Another ex-IBMer, Tivoli CEO Frank Moss, now returns
to the company at the senior executive level, as head of a yet-to-be-named
division that will include more than 300 developers working in Research
Triangle Park and Rome, Italy, along with Tivoli's Austin-based developers.
On the sales side, Alex Kuli, an ex-IBMer who headed Tivoli's sales staff
of about 70, will oversee a systems management sales force of more than
300. According to Neath of Tivoli, IBM plans to nearly double this sales
force by the end of this year.
Carping and Competing
With a beefed-up sales force and an open object-oriented framework, IBM
suddenly becomes a serious contender for MIS mind-share at a time when many
IS managers are struggling to manage a growing number of heterogeneous systems
without increasing their staffs. Computer Associates responded to this competition
by running full-page ads in computer industry trade publications with headlines
that refer to Tivoli's loss of independence and by openly soliciting TME
developers with job inquiries.
In response, Tivoli distributed an open letter to customers and industry
colleagues during UniForum '96 in February. In it, Moss refutes the idea
that TME is on the verge of becoming less open. No preference will be given
to IBM platforms, he argues, promising that Tivoli will continue to expand
its partnership relationships, for example, by working with Oracle Corp.
to support a TME-based Oracle management product. Geschickter of Hurwitz
Consulting reasons that customers need not worry about TME developers focusing
resources on IBM platforms to the exclusion of others. If IBM were to take
a closed system approach, it would sabotage the value of its own asset,
Tivoli will compete head-to-head with HP in the race to offer the most platforms,
to sign up new partners for its framework and to provide a deeper level
of integration for previously integrated tools. TME currently supports 20
platforms, including 14 varieties of Unix as well as Microsoft Windows NT,
which has been available for over a year. In March, HP announced that it
would introduce NT-based agents for its OpenView systems, network, application
and database management solutions during the first half of 1996. On the
Unix side, OpenView supports eight platforms, including AIX.
Application performance management is another area in which system management
framework vendors compete for mind-share and alliances. In 1995 Tivoli launched
its application management strategy and introduced a management solution
for R/3, the integrated client/server suite from SAP America of Philadelphia.
HP says that a version of OpenView IT/Operations (formerly called HP OpenView
Network Node Manager) for managing R/3 will be available in the second half
of 1996. Action Request System, a client/server help desk application from
Remedy of Mountain View, CA, which has been integrated with HP OpenView
for over two years, will increase the depth of its integration with OpenView
and also be integrated with TME this year.
Competitive pressures produce strange bedfellows. IBM's intent to absorb
Tivoli invites comparison to another recent acquisition. If the acquisition
proceeds as planned, systems management will become one of five software
groups managed by John Thompson, an IBM senior vice president. The groupware
division, led by Lotus Notes, is another of the five. Like Lotus, Tivoli
will continue to keep its brand name. The reorganized products group that
Tivoli will join will be Tivoli-centric, says Neath.
A more skeptical view of these acquisitions is that IBM is buying time-to-market
and recovering from misbegotten attempts at creating strategies for distributed
open systems. "IBM is going outside the company to clean up its systems
management act, the way it did last year [by acquiring Lotus] to get its
groupware act together," says Geschickter. Time will tell how well
the money has been spent. -Peggy King
Plan 9 Draws a Crowd
One of the pleasant surprises during the UniForum '96 Conference in February
was the enthusiastic response to the extended session "An Evening with
Plan 9." At the gathering, an audience of more than 500 serious open
systems professionals heard the latest available information on the new
operating system, which originated in the famed AT&T Bell Labs. Two of the
principal developers of Plan 9--Dennis Ritchie, the Unix legend, and David
Presotto--addressed the crowd in the packed room.
Plan 9 is finely tuned for distributed computing and consists of three parts
that can be installed on separate systems: a network file server, an applications
server and a client. It can run on systems as low-powered as the Intel 386
processor, as well as on Sparc and Mips RISC chips and the Motorola 68020
found in some Macintoshes. It can read Unix files and supports the Posix
and ISO C environments. Plan 9 also features a windowing system and support
for machines that use multiple processors.
An air of expectation in the room gathered around Ritchie's celebrity in
open systems circles. Both he and Presotto are now part of Lucent Technologies
in Murray Hill, NJ, the systems and technology company formed after AT&T's
most recent restructuring. Aided by the research and development resources
available from Bell Labs, Lucent's charter is to design, build and deliver
public and private networks, communications systems and software, consumer
and business telephone systems, and microelectronic components--an impressively
Ritchie said that Plan 9 began as a research project prompted by three thoughts:
that Unix is now "boring"; that networks and graphics capabilities
are firmly established aspects of computing; and that traditional computers--mainframes,
minicomputers, workstations and PCs--are not the final answer to evolving
needs. The key ideas in Plan 9, he said, are that all objects are files
named in the file system name space (this category includes real files,
devices, networks, processes and services); that all operations on files
can be turned into network messages to a file system server; and that what
the file name space applications see is dynamically and individually bound
to running processes. Therefore, the applications are "unconscious"
of where they are actually running, and services are provided in a uniform
According to Ritchie, one of the principal concepts driving the development
of Plan 9 is "extreme platform independence," under which all
platforms access a common file system. Different machines can then be connected
through various kinds of networks and yet be dealt with in a common way.
The major hardware components for Plan 9 don't include the usual systems.
Rather, they are
terminals equivalent to diskless workstations, possibly with file caches;
CPU servers to which large computations can be exported and special services
can be imported;
and fast network connect servers with slower networks connecting terminals.
Ritchie stressed that Plan 9 research is directed at structure, not features.
Some of his remarks certainly appealed to purists in the audience. "We
are not in the general operating system business; we are producing ideas
and examples," he asserted. He added that Unix, Windows and Windows
NT are "hard to compete with" and that it really isn't feasible
in the current industry environment to create another general-purpose commercial
OS. He cited the tribulations of OS/2 Warp as proof. Instead, Ritchie expressed
hope that Plan 9 will be adopted for commercial applications such as set-top
boxes, for which its small size might be appropriate. It has a 300K kernel
program as opposed to 500K to 1.5MB for commercial Unix.
Attack of the Java Killer?
In response to a question at the end of his presentation, Ritchie alluded
to a project of "looking into Plan 9 technology" to create a transportable
operating system and programming language. This was, of course, a reference
to Bell Labs' supposed rival to Sun Microsystems' Java, code-named Inferno.
In statements subsequent to his appearance at UniForum, Ritchie called the
concept behind Java "compelling," adding that any proposed AT&T version
would be designed to be useful on a variety of machines, perhaps even including
high-tech future televisions.
It is not clear now exactly how far work has progressed on Inferno, but
the project is apparently of such importance that most of the Plan 9 staff
has been diverted to its development. It is also unclear whether any decision
has been made to actually bring it out as a product. However, industry analysts
say that there is interest in Inferno at the highest levels of management