The mainframe is celebrating its 40th birthday, writes Bloor Research president Robin Bloor. It must be said it has stood the test of time like very few other things in the IT world. History will tell you that Tom Watson Jr. made a $5bn bet on the mainframe (then referred to as the System 360 or simply s/360) which IBM claims to be the largest privately-financed commercial project ever.
Do you remember? - quite probably not. Maybe you weren't even born. But just to make it clear - in those days it was all punch card and paper tape. It was batch processing only and any major computer installation had an office full of female (and sometimes very fearsome) punch card operators. When you submitted a program for testing, it was a batch of cards which were kept together by a rubber band and if you dropped them accidentally, sorting them out was no simple matter.
It wasn't long before on-line processing emerged with CICS - which also still exists and runs mission-critical systems, but in those days was usually referred to as a TP monitor. The following generation of computing (the minicomputers) had operating systems that were designed for TP but MVS was originally designed for batch, so you had to put CICS on top. It ran in its own partition.
Partition? Yes indeed the partitioning of the fundamental computing resources in the mainframe was a necessary foundation for managing resources. And guess what? - the wheel is being reinvented now, with the whole idea of flexible partitions for servers (with VMWare, On Demand, etc) - servers that are more powerful and more poorly managed than the original mainframes ever dreamed of being. This flexible partitioning rescued IBM's mainframe from a slow death because it proved to be an excellent and very economic platform for Linux.
The mainframe oversaw the invention and birth of many things. High availability? Yes indeed. It was not invented anywhere else. The idea of database was born on the mainframe, as was the idea of a 4GL, as was (believe it or not) word processing and voice recognition. Also - and here we can pause for a smile - system management was born on the mainframe and lived there and never died, but it is now being "born again" for the corporate network.
Indeed just about everything that makes up the whole "On Demand" bandwagon that everyone and his dog has now jumped on, is a mapping of mainframe computing to a computer network. The only exception I can think of is "fault tolerant computing". For that, Tandem Computing and Jimmy Treybig can take a bow - but to be honest everyone referred to Tandem as a mainframe anyway.