Section 29.6. CONCLUSION

29.6. CONCLUSION

This chapter has presented some examples of crosscutting concerns in database systems. Two object database evolution systems, SADES and AspOEv, have been used as a basis for discussing how such concerns can be effectively modularized with AOP techniques. Such modularization results in improved customizability, extensibility, and maintainability. A range of crosscutting concerns exists in database systems, and the discussion in this chapter has barely scratched the surface. Interested readers are referred to [28] for a more detailed discussion on the synergy between AOP and database systems and how the two technologies can offer mutual benefits.

Like all new technologies, the introduction of aspect-oriented concepts in database systems must be approached with caution and in an incremental fashion. One possible approach is to initially employ aspects for internal system optimizations. After such optimizations have been successfully implemented, the concepts can be exposed to the application programmer. This approach was effective during the development of SADES, where AOP techniques were initially employed internally for reducing the impact of schema changes before exposing the aspect language to the application programmers for customizing the instance adaptation mechanism. Another factor to be taken into account when introducing aspect-oriented techniques in database systems is the dependencies among the various crosscutting concerns being modularized. One must identify the concerns that build on the functionality offered by other concerns. The latter can then be implemented in lower system layers, exposing their functionality to the former that reside in higher system layers [25].

Commercial acceptance of the concepts presented in this chapter largely depends on the efficiency and correctness of weaving mechanisms. Therefore, the development of techniques for efficient weaving and its verification requires extensive research. Nevertheless, aspect-oriented concepts hold great promise for database systems and are bound to be increasingly adopted in this domain.