Twenty Years of Jobbers
Chapter X - The Interface Problem
A book in progress by Bill Scott
(c) Copr 1998 ScotSystems Inc All Rights Reserved
Many businesses feel that one vendor should be able to provide all of the data processing needs in one package. If the facts be known, just as every tool has its use, there may be several different
components that make up a complete data processing package, the core of which is a strong and reliable back-office accounting system - the very heart of the business. Peripheral systems, such as cash registers, scanners and
card readers must be kept separate so that they may be switched without "infecting" the heart of the system with problems created by change. In other words, the back-office system has to be strong enough to allow
itself to be fed by different peripheral systems as these systems evolve. This can only be possible if the back office system is inter-dependent of these alien devices and makes itself accessible to them through a common
Much has been written about the so-called "open-systems" issue but the meaning of "open-systems" has been misused over the past few years. A true "Open-System" bears allegiance to
no single operating system such as Windows/95, NT, OS/400 or Unix, because doing so would be an oxymoron - defying the very meaning of "open-systems". The true meaning of an "open-system" is one in which
several "closed-systems" can be integrated by way of standard interfaces.
What is a standard interface? Well that's the problem isn't it? We are able to switch refrigerators in our homes because the
"open-system" interface is a common electrical outlet. Do refrigerators communicate with other devices on the same circuit? While there may be no direct connection between the refrigerator and the electric range,
there is one device that all systems connected to an electrical circuit must communicate. The "electric service" panel connects all systems to the power grid. The circuit breaker is an interface policeman located in
the electric service panel that insures all devices connected to the power grid don't overload the circuits. A common interface in electrical circuits is possible because the appliance industry matured to an agreed upon set of
standards or rules. It's not necessary to buy a new electric range when you replace your refrigerator. Why should it be necessary to buy a new scanner when you change your cash register? The only answer must be that the
computer industry hasn't matured to the point where standards can be enforced.
At some point in the history of electrical circuits, it became clear that a common interface into the power grid system had been chosen. Then
the various manufacturers were governed by the strict self-imposed rules dictated by the interface. If they refused to standardize their circuits to match the interface to the power grid, they were excluded from the game. Such
a time has come to the automation industry.
It will be a shame if all this stuff about the Year 2000 overshadows the need for a common interface in data processing, but I'm afraid that is exactly what is happening. If we
don't develop a common interface between trading partners, we may be in real trouble quite soon.
Just as there was a moment of enlightenment when we became aware of a common interface for electrical circuits, we have
long-since reached that threshold in the world of data processing. Once thought to be EDI (Electronic Data Interchange), the standard for a common interface in data processing will be TCP/IP and SQL. Anything else will be
localized to the individual companies who may have plans to utilize other interface options internally. By revising our systems to be compliant with TCP/IP and SQL, we will be doing what is necessary to insure that we will be
able to conduct business electronically with our trading partners.
What's the danger if you don't do this? I believe that without some type of electronic interface to our trading partners, we will not be able to conduct
business in the 21st century. It's that simple. Not only will our trading partners demand it, competition will be such that not having it will be tantamount to operating a business without water and electricity. It just won't
IBM has shown the way by giving us the AS/400 and DB/2. Other vendors may use whatever methods they chose to connect with the necessary TCP/IP interfaces, but they must all be SQL compatible or the cost to
reformat the data for connectivity will be cost prohibitive.
It may be important to note that it may not be sufficient to simply connect with our trading partners using TCP/IP. If we are going to hold costs down, our
internal connections must be TCP/IP as well. This means that cash registers, pump controllers, scanners and computers will have to connect via TCP/IP. Therefore if our internal systems are TCP/IP compliant, then equipment can
be switched out easily with other TCP/IP compliant equipment. This will greatly reduce the expensive and tedious problems of dealing with system upgrades. Technology will continue to change. A common standardized interface
between equipment will buffer out the effects of change and give us time to manage change on our own terms.
Do providers of POS equipment truly believe that their proprietary choice for connectivity will become the
de-facto standard? There is a possibility that their engineers see common interfaces as the end to their individuality. That they will no longer receive the high salaries they have learned to depend upon. I believe that it is
only through standards that we can advance past the "guinea pig" stage of research and development and into a mature market where we can all prosper. NACS is correct. Once the obstacle of "standards" has
been overcome, then and only then will we see real progress being made in automation.
After the Year 2000 problem is long forgotten, the issue of trading electronically will be our next huge obstacle to overcome.
Companies that are able to prepare for this hurtle early, stand a much better chance of surviving in the next millenium.