System Integration

Dedicated to the dissemination of System Integration information

Design for System Integration

For the most efficient System Integration (S.I.) the system should be designed with S.I. in mind.  I don’t think anyone would disagree with this but, on many occasions, it does not happen.

 

As milestones loom all thoughts turn inwards.  S.I. is a long way off, the looming milestones do not include the needs of S.I., as far as the milestones are concerned the needs of S.I. are irrelevant.

 

The problems are only discovered during the S.I. phase, you now find that you cannot get any information out of the system, that the interfaces are complex to monitor and understand, S.I. is an almost impossible task.

 

It would have been better to include the S.I. requirements in the milestones, define what S.I. needs to be successful and ensure that they are built in from the start and are included in the milestones.

 

A problem with large systems is that the time from a blank sheet of paper to the S.I. phase can quite often be years.  The hardware for a system will be defined fairly early in the project.  Any test interfaces will be designed with current test equipment in mind, i.e. by the time of the S.I. phase the test equipment being used will be out of date, or at the minimum it will not be the best available. 

 

There is currently research being carried out into Platform Independent Models (P.I.M.) but this is only for the software component of a system, the idea is that the software is designed and it is only at the last moment that the software is loaded onto a system.  As the software design is platform independent the hardware can be defined at a later stage in the life cycle and the intellectual property of the software design has a longer life, i.e. it can be reused.