The growing size of software models poses significant scalability challenges. Amongst these challenges is the execution time of queries and transformations. In many cases, model management programs are (or can be) expressed as chains and combinations of core fundamental operations. Most of these operations are pure functions, making them amenable to parallelisation, lazy evaluation and short-circuiting. In this paper we show how all three of these optimisations can be combined in the context of Epsilon: an OCL-inspired family of model management languages. We compare our solutions with both interpreted and compiled OCL as well as hand-written Java code. Our experiments show a significant improvement in the performance of queries, especially on large models.
- Query performance