The British Computer Society journal on the embedded problem

greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread

The British Computer Society on the Embedded Problem

http://www.bcs.org.uk/publicat/ebull/july99/y2kembed.htm

The Computer Bulletin Year 2000

July 1999

Embedded problem

Embedded systems are the great unknown in the year 2000 issue: it is impossible to find them all and impossible to know if they will be affected in any case. Steve Goodwin outlines the problem and the risk

Awareness of the year 2000 problem is generally much higher for business systems than for embedded systems, and when it comes to investigating and fixing the problem, there are even greater differences.

Embedded systems are pieces of electronic equipment used almost everywhere: in manufacturing and process industries, buildings, testing and diagnostics, telecommunications, medical diagnostics, construction, transport, and so on. There are billions of them and they are everywhere - and harder to find than business systems.

They may be microprocessors in temperature control units, automatic flow control valves, gas sensing devices and so on. Lifts and security alarms may have them.

Although there are billions of embedded systems out there - and in theory all of them should be considered - the great majority will turn out either to have no date or time functionality at all, or to be compliant.

The trouble is that although the incidence of year 2000 problems in embedded systems has been found to be relatively low, the impact of the problems that there are may threaten businesses.

It is useful to consider embedded systems in three areas: microcontrollers, microprocessor-based systems, and large scale systems.

Microcontrollers, such as those found in door chimes, microwave ovens, fridges and so on, have minimal memory and cannot be programmed by users - and there are millions of them, so it is impossible to take an inventory of them. However, the likelihood of them having a date problem is of the order of 0.001%, so in most cases they will not be high priority.

Microprocessor-based systems on the other hand, although they still have minimal memory, can often be programmed to some extent. Taking an inventory of these is tedious, but it should be done, as problem rates are put at between 1% and 6%.

Large-scale systems - sometimes described as 'things with discs' - involve a computer of some sort somewhere, so all the year 2000 problems that afflict business systems may affect them, too. This means the number with problems is significantly higher than the number of microprocessor-based systems.

Both hardware and software must be compliant: one without the other is of limited use.

Experience so far shows that the number of problems found tends to vary with the type of business and also that there seem to be more problems arising as a result of leap year errors than from the move from 1999 to 2000.

The lowest level of problems has been found in simple manufacturing industries: 2% or less. There have been fairly low levels in building services - up to about 6% for complex buildings - and a very wide spread of higher values in the more complex manufacturing and process industries.

So what should we be doing?

The IEE has been concentrating on the embedded systems side of the problem, while our colleagues in the BCS focus on business systems, and, like the BCS and others for business systems, promotes an approach based on risk management. The main steps are making an inventory; assessing the business impact; compliance assessment and testing; risk assessment; and starting the action needed.

An IEE Technical Guideline, Year 2000 Problems in Machinery - Managing the Risk in Smaller Companies, explains all this in detail. This was produced last year and is still very relevant - but there is now much less time to do everything properly. Companies that have not yet started will have to accept a higher level of risk and cut some corners. As the deadline grows ever closer, it becomes more and more important to set priorities for the investigative and remedial work.

The government's Action 2000 body, with help from the IEE, is collating facts about potential year 2000 failures in a wide range of embedded systems. This information on the areas at greatest risk should help companies focus their efforts. In particular, they should concentrate on the critical business systems and on the higher risk embedded systems.

It must be stressed, however, that all year 2000 work is an exercise in risk management: no organisation can guarantee 100% compliance. Time is of the essence: this is one deadline that cannot slip. Many companies will need help - how much help will vary - and good help is becoming hard to find.

Unfortunately, too many unqualified and inexperienced people are jumping on the bandwagon and offering to solve all the problems. There are, however, excellent consultancies of all sizes, from sole practitioners to the largest firms: it is important to make sure that the chosen expert really knows his or her stuff.

The year 2000 problem is not going to go away - not for a year or so, anyway. It is important to ignore the hysteria and use common sense and engineering expertise to mitigate the effects as far as possible.

Remember: there is a problem, but most of it is surmountable - if we keep working at it.

Spread the word!

Steve Goodwin is managing partner of The Houndscroft Partnership and chairman of the IEE [Institution of Electrical Engineers] Millennium Embedded Systems Working Group. He is a Fellow of the IEE and a Chartered Engineer.

-- Old Git (anon@spamproblems.com), September 10, 1999


Moderation questions? read the FAQ