If your were implementing a system for a real company then once you had created it you would need to install it at the company's premises. If your new system was replacing an older existing system you would need to decide how to
carry out the changeover from the old to the new system. Data that was stored in the old system may need to be transferred onto the new system (possibly using a standard file format).
If the system you created carried out an essential task for a company then they may be reluctant to simply give up using their old system and replace it with a new one until they were convinced that the new system worked properly.
One of four different changeover methods might be followed :
| Method | Description
|
| Direct
| The old system is scrapped immediately and completely replaced by the new system. This requires minimum effort, but if there are any problems with the new system the data in the old system will not
be up-to-date so it will be hard to return to it. Return to the old system would be harder still if the hardware it used was removed. This changeover method is most frequently used on systems that are not essential to the running of a business.
|
| Parallel
| Until it is decided that the new system works correctly the new and old system are used together. Every time a transaction occurs (e.g. a sale) it is entered into both systems. This is very time consuming
but if there are any problems with the new system they can be easily identified by comparing the output of the two systems. It is also easy to revert back to using the old system if the new system does not work as all the data in
it will be up-to-date.
|
| Phased
| The new system is tried out on a small scale to see if it works. For example a supermarket might introduce a new stock control system at just its Bury branch. At the other branches the old system remains in use.
When the supermarket is happy the new system works the old system will be gradually replaced at its other stores with
the new system. Any problems that occur would only affect the store(s) that the new system was being tried at.
|
| Pilot
| The new system is run alongside the old system, but only processes some of
the data. For example a supermarket might try out a new stock control system on one till in
a
store, with all sales at this till being recorded by both the old and new systems. Sales at this till may take
longer, but the amount of effort required is less than that for a parallel changeover. The results produced by the two systems can be compared to see if the new system appears to work. As the new system is only being used with
a small amount of data, problems that might result from processing a realistically large volume of data might not be noticed.
|
