Conclusion

The evaluation of the methodology for real cases demonstrates that the method is a tool allowing us to increase the availability for software solutions whose architecture can be split into components, insofar as the physical components can be abstracted as virtual components and where there are mechanisms internal to the logic components allowing selection of valid duplicates.

Although the method of analysis enables one to increase the availability of a software solution, other factors must be taken into consideration.  For example, the customer of the solution can require high availability for only some components of the solution.  The methodology could be adapted to incorporate the notion of necessity for software components; the availability analysis would then be weighted by the logic components as well as by the virtual components of the solution.

Also, increasing the availability of a component can have a financial impact on the implementation of the solution.  This new variable could also be added to the analysis of bottlenecks, and could guide the prioritization of software components that require improved availability.  The result of the analysis would then be a derivative of the one of availability analysis.

Finally, the scope of the method of analysis focuses on duplication as an availability improvement mechanism.  In certain contexts, replication could be nuanced to take advantage, for example, of a specific implementation of a virtual component.