Information Technology Reference
In-Depth Information
representatives, future CDC ALS project manager Fred Laccabue, remembers discus-
sions at Hanscomb Field far differently.
14
He and other CDC officials believed the
plan had a major design error: that the Air Force was writing the central control sys-
tem using COBOL rather than machine code. COBOL could not approach the effi-
ciency of assembly language (native language of a particular computer). The use of
COBOL was to try to provide operating system-like functionality, such as job sched-
uling, to ALS. The small CDC team, including Laccabue, expressed a few general
concerns and requested a subsequent private meeting with the developers of the RFP
in 1969 [18]. There, the CDC team explained in detail what they saw as problematic
elements (including the use of COBOL) in the RFP specifications that would be “real
impediments to achieving success for the program.” [18] In Laccabue's words, the
CDC team was “not so politely rebuffed.” [18] They were told that the Air Force had
“employed many experts in the computer field” regarding the central control system
and unified database, and were “absolutely confident they were going down the
correct path.” [18]
Elliott correctly acknowledges that CDC had emerged as the number two com-
puter firm (behind only IBM) in profitability by the late 1960s, but he generally
presents CDC in an unfavorable light. He suggests that CDC was a relatively new,
untested, and risky company that “almost went bankrupt before its first computer was
delivered.” [2] This vision of CDC contrasts sharply with what CDC was when it bid
for ALS. By the late 1960s, CDC had emerged as the world leader in supercomputers,
owing in large part to the skill of arguably the most gifted computer design engineer
alive, Seymour Cray. It had established a profitable computer peripherals division and
was thriving as it concomitantly sought to extend its success in scientific computing
and build its capabilities in business data processing hardware, software, and services
[19]. Though Elliott mentions a qualifying ALS benchmark test for which specifica-
tions were distributed in July 1969, he fails to disclose that CDC was the only firm to
pass this ALS test in its first, and originally its only planned, incarnation. IBM (using
dual System 360/67 computers), Sperry Univac, RCA, and Burroughs all failed [2].
Despite the Air Force's overarching goal to move quickly with ALS (after all, the
nation was at war in Vietnam), it decided to delay the contract award process roughly
six months to allow other computer firms to qualify to bid. This extension upset lead-
ers at CDC [18]. Control Data had invested heavily to set up a huge complex at their
Sunnyvale, California facility to house the mass storage disk drives and mainframe
computers to meet the benchmark. In fact, they had computers and mass storage suffi-
cient to run the test transactions in roughly twelve minutes, when the benchmark was
sixteen minutes [18]. The dozen or so Air Force officials had already watched the
other firms fail, and initially, were elated to see CDC's results. Top Control Data offi-
cials believed extending the testing for a half year for a second round was just a
means to drive down the end price for the contract [18].
Ultimately, IBM dropped out of the competition due to conflicts regarding liability
specifications, and Sperry Univac (having passed on its second try) was the only firm
14
Evidence indicates not only CDC, but other potential vendors cautioned “that there was a
serious question on the availability of adequate software and that it might be beyond the
state-of-the-art.” [6]
Search WWH ::
Custom Search