System/360: The big gamble
IBM was at the top of the computer heap in the early 1960s, but all was not well. There were many competitors selling their own systems, claiming to be faster, better and cheaper (as the old joke about IBM goes, "You can buy better, but you won't pay more"). IBM itself had three competing, non-compatible computer divisions and flattening revenues. Worsening the problem was the incompatibility between upgrades of the same system family forcing customers to continuously buy new software and peripherals. IBM executive management decided it was time to take a risk.
The primary goals for the new system were compatibility, both in software and hardware, and a smooth upgrade path. There were many problems along the way, some of them technical.
The technical issues included using a relatively new electronic technique called Solid Logic Technology (SLT), which was something between individual transistors and an integrated circuit. Not content just to use the new technology, IBM decided to manufacture it as well, which meant building and manning new factories. Along with the new processor, IBM decided to create a line of peripherals compatible with the new system.
The new system needed new software as well. Programmers were not only tasked with writing an operating system (OS) that worked on all System/360 incarnations, they had to develop one of the first OSes to support multiprogramming and interactive processing. It's telling that software ended up being the biggest wedge in the System/360 development budget, ballooning from the original $30 million to over $500 million. In the end, the programmers created three different operating systems to cover the 360 spectrum: Basic Operating System (BOS), Tape Operating System (TOS) and Disk Operating System (DOS).
Then there were the human factors. As mentioned above, IBM had to stop the upgrades already in progress for the three old computer lines. And, as we IT workers know, opinions are like pocket protectors, every technician has one. This meant IBM had to convince hundreds of technicians with personal and professional stakes in the old machines to drop everything to work on System/360. It took a lot of arm twisting by forceful personalities to get everyone on board before the first the first circuit was drawn. Even after the announcement there was still dissent throughout the company and many clucking tongues.
The results of the mainframe
Despite the ordered chaos, IBM announced System/360 on April 7,1964. The system must have been well received, as IBM took orders for around 2,000 systems in the first eight weeks. After sales came the hard part, when IBM had to build the factories to build the machines and hire programmers to write the code to run the system. It took another two years for System/360 to be considered successful. By 1966 there were between 7,000 and 8,000 systems installed, creating $4 billion in revenue. Best of all, IBM delivered on its compatibility promise.
The paper mentions that IBM made the original hardware specifications and source code available to everyone. In a wonderful foreshadowing of open source, the availability of this information caused a huge burst of customer- and vendor-driven innovation, something I think IBM could use now.
I think it's also important to note there were some things the System/360 couldn't do. For instance, IBM realized the mainframe left a big hole in the small and medium-sized market, which was eagerly filled by the competition. IBM bridged that gap by creating other lines of smaller (and incompatible) computers, such as System/36 and AS/400.
In the end, System/360 did not bring computing to the masses, midwife the Internet or change the music industry, but its influence is considerable. System/360 was the first computer to use 8-bit bytes instead of the 6-bit bytes that were the standard at the time. The IBM mainframe may not have been the first to implement some technologies, but it is certainly where they were perfected. The list includes things like virtual storage, virtual machines, resource security, relational databases, system integrity through storage keys and authorized instructions, print spooling, job scheduling, online processing, multi-programming and multi-processing. Then there are other technologies that other platforms have yet to implement, including granular system maintenance, workload management and rational systems management. I also think it's significant that IBM can proudly point to 45 years of unbroken compatibility. You may not want to run that program that hasn't been assembled since 1974, but, in a pinch, you could. Would you try that on any other platform?
Forty-five years later, the future of the mainframe is a little cloudy as other hardware gets cheaper and Windows and Unix system management gets easier. But, even if the mainframe is destined for the bit bucket of history, future generations should remember its contributions and think about whether the computer world would be where it is today if IBM hadn't taken the big gamble.
ABOUT THE AUTHOR: For 24 years, Robert Crawford has worked off and on as a CICS systems programmer. He is experienced in debugging and tuning applications and has written in COBOL, Assembler and C++ using VSAM, DLI and DB2.
What did you think of this feature? Write to SearchDataCenter.com's Matt Stansberry about your data center concerns at firstname.lastname@example.org.