Discussions‎ > ‎

CONCOMP: Conversational Use of Computers

NECESSITY, THE MOTHER OF INVENTION: MTS and the DATA CONCENTRATOR

posted Jun 9, 2014, 5:37 AM by Jeff Ogden

From page 1 of the U-M Computing Center Newsletter, Volume 1, Number 4, 8 March 1971:

The Computing Center's IBM System/360 Model 67 was
delivered in January 1967, and since that time the hardware
has proven itself to be as versatile as we had anticipated.
However, IBM's operating system for the 360/67, known as
TSS (Timesharing System), was not available on that date,
and, when finally available, its performance was not
adequate to discharge effectively the demands placed upon
it by the great diversity and large number of University
users. In addition, there was not available as standard IBM
equipment a "terminal controller" which would permit
effective communication between the central computer and
interactive terminals having widely varying performance
characteristics.

These two shortcomings placed a particularly awesome
burden on the Concomp Project: Research in Conversa-
tional Use of Computers, a multimillion dollar project
funded by the Advanced Research Projects Agency of the
Department of Defense to study the wide range of topics
implied by its title. This Project (recently terminated)
began in August 1965, and was predicated on the avail-
ability of both TSS and the terminal controller.

Thus the Computing Center and the Concomp Project
found themselves involved in two large research tasks that
they had not anticipated: (1) the development of an oper-
ating system for the central computing facility that would
support any type of effective man-machine interaction, and
(2) the development of an effective hardware interface to
support a wide range of terminal devices.

The successful resolution of the first problem by the
Computing Center resulted in the operating system known
as MTS (Michigan Terminal System). Mr. Michael
Alexander of the senior staff was its chief "architect." MTS
now serves more than 9,000 active users, and is a single
unified system for both interactive terminal use and batch
processing. A typical demand peak finds fifty to sixty
active terminals and four to six batch streams operating
concurrently.

The success of MTS is attested to by its adoption by a
number of other universities: Wayne State University in the
U.S.A., Newcastle-upon-Tyne in the U.K., Grenoble
(France), and the universities of Alberta and British
Columbia (Canada).

Remote terminals using MTS may range from the
ultimate low cost and simplicity of the pushbutton, or
Touch-Tone, telephone, through the numerous low-speed
alphanumeric terminals (both cathode-ray-tube and impact
printers) and storage-tube graphic displays, to the more
versatile remote graphics terminals. Moveover, the facilities
for conversational conputation provided under MTS are suf-
ficiently general to permit their extension to many other
areas not originally proposed-for example, MTS provides
unusually flexible remote-job-entry facilities as well as
interaction between the central computer and remote mini-
computers in a variety of laboratory digital/analog control
systems.

The lack of adequate terminal controller hardware for
interfacing diverse remote-access terminals with the System
360 hardware led to the design, development, and fabrica-
tion under the Concomp Project of a very generalized data
communications terminal controller known as the Data
Concentrator. It was inspired and implemented by Mr.
David Mills of the Computing Center's senior staff. The
generality of the Data Concentrator stems from the fact
that it is programmable, its principal component being a
small general-purpose computer, the DEC PDP-8. This
means that its designers can program the Data Concentrator
to accept a variety of terminal of different manufacture and
different operating characteristics. This fact, coupled with
other characteristics of the Data Concentrator, has made it
possible for the Computing Center to respond to the
evolving remote-terminal requirements of its user com-
munity.

In addition to adaptability, the Data Concentrator is
characterized by communications protocols which facilitate
interactive use of MTS. The development of the Data Con-
centrator's communications protocol and general organiza-
tion served in part as the basis for subsequent development
of computer networks, such as the ARPA network and the
MERIT Computer Network; and the hardware designs have
influenced the construction of several contemporary
machines which will be used to interconnect the major
computers in the MERIT network. According to F. H.
Westervelt, formerly Associate Director of the UM Com-
puting Center and now Director of Wayne State's Computer
and Data Processing Center, "The flexibility of the Data
Concentrator in responding to a dynamically changing en-
vironment of line speeds, code frames, code conversions,
line controls, and device-support requirements is still un-
matched in any commercially available equipment."

The success of the Data Concentrator, coupled with its
ever increasing use by greater numbers and varieties of
terminals, led to the recent decision to begin construction
of a second Data Concentrator, this time based on a
PDP-8E and a PDP-11 computer. These are machines of a
newer generation than the old PDP-8, and they are ex-
pected to accommodate more adequately the anticipated
increased demands for Data Concentrator facilities.

CONCOMP papers and reports online

posted May 21, 2014, 5:33 AM by Jeff Ogden

The following papers and reports are available online:

CONCOMP Project Publications, 6 September 1971

posted May 20, 2014, 6:17 PM by Jeff Ogden   [ updated Jun 9, 2014, 9:41 AM ]

1-3 of 3