There are three methods of legacy modernization
- Build again from scratch
- Build again, but piecemeal
- Convert, using PCM
Build again from scratch
This is the most difficult but not the most expensive method. It is difficult because you must replace a legacy system – which has been evolving for many years – with a more modern system which has more functionality but which includes, at the same time, all the functionality of the old system, in the shortest possible time. In most cases this takes 5 times as long as planned, and is at least 5 times more expensive than budgeted for – that is if you ever succeed!
Build again, piecemeal
With this method you rebuild the new system module by module. With each phase you make an interface between the old legacy system and the new module. This method works well but is more expensive than building the system from scratch. This is logical, because you also need to build and maintain the interfaces, so it takes far longer to replace the legacy system.
Other disadvantages:
- Slow performance. The old system had integrated data transfer; now it is done in two separate systems.
- Overheads. The developer team must synchronize their efforts.
- Mutual dependency between development teams. They often have to wait for each other. This mutual dependency is costly in time for management due to detailed and synchronized planning
At the present moment this piecemeal type of legacy modernization is mostly done using Web Services, the method used being SOA. A few years ago this was mainly hype, but the hype is slowly transforming into a more down to earth assessment about what is possible and what is not.
Parallel Continuous Migration.
Parallel Continuous Migration is our method and one which we are very proud of. It is the quickest, the least costly and we think, the only method to safely replace legacy systems.
