Demystifying enterprise IT, part 1: SOA - Interview with John Osborn (iTWire podcast)

Tony Austin
Friday, 10 October 2008 05:57

IT Industry - Market

Microsoft Office 365

When I started with computing, in the late 1960s, and architect was understood to be somebody who designed  homes or buildings. But things were beginning to change. As Wikipedia quite rightly points out the IBM System/360 was "the first family of computers making a clear distinction between architecture and implementation."

The implementation was the actual nuts and bolts, so to speak, of the various S/360 machines, while the architecture reflected the overall abstract design of the processor family.

Indeed, the "360" part of the System/360 family name was intended to stress that the architecture was a general purpose one that spanned the "full 360 degree compass" or the "full spectrum" of computing needs, covering both scientific computing (such as floating point arithmetic) and commercial computing or "data processing" (such as decimal arithmetic and character manipulation).

Wikipedia goes on to explain computer architecture as "the conceptual design and fundamental operational structure of a computer system."

It is a blueprint and functional description of requirements (especially speeds and interconnections) and design implementations for the various parts of a computer — focusing largely on the way by which the central processing unit (CPU) performs internally and accesses addresses in memory."

Moving away from hardware, in those days, there weren't many software standards to speak of. There had been a few versions of computer languages such as FORTRAN, and compilers were amongst the first software where the need for standardization was recognized.

In 1974, one of the first large-scale software architectures was released, also by IBM. This was Systems Network Architecture (SNA). It was a proprietary networking architecture, brought out by IBM as the then unchallenged leader of the IT industry to fill a need.

You rarely hear the name SNA mentioned today, even though it pioneered layered architectures, in which functions are compartmentalized into independent (but interdependent) layers to provide optimal flexibility and reusability.

SNA had six layers (I'm pretty sure), but it was proprietary and along came Open Systems Interconnection (OSI), a joint ISO and TU-T standard for computer networks and communication protocols. The OSI model was also a layered architecture, similar to SNA but with an additional seventh layer.

Both SNA and OSI were very much hierarchical models, centered on to match the centralized organization with a focus on mainframes and static networking requirements. These days, you rarely hear a mention of either SNA or OSI, so much for two of the glories of IT antiquity.

Oh, the many hours, days, weeks that I spent at IBM learning the innards of SNA architecture and product implementations: VTAM, NCP, APPC (alias LU 6.2), and much more. All this knowledge of no use to me any more, except as nostalgic feel-good memories.

In the late 1960s and early 1970s the Defense Advanced Research Projects Agency (DARPA, part of the US Department of Defense) built its ARPANET network and was moving in another direction, that of networking flexibility and resiliency (think nuclear warfare, where centralization is anathema). DARPA's work eventually led to the specification of the Internet Protocol Suite, commonly referred to as TCP/IP, or sometimes as just TCP or even IP.

Crikey, mate. We haven't left the 1970s behind and already we've covered enough complex architectures, standards, protocols and acronyms to drive you batty. Let's get out of the dark ages and jump to the present time, shall we?

PLEASE READ ON...