资 源 简 介
This homomorphism-based framework provides higher level programming interfaces for wrapping the Google’s MapReduce and try to make programming with MapReduce becoming easier. The key idea is to build a homomorphism layer upon MapReduce, which can automatically generate homo- morphisms from user’s input sequential programs and implement the generated homo- morphisms by MapReduce. This homomorphism layer not only makes the analysis of parallelism easier, but also guarantees the generated MapReduce-programs to be with good parallelism and scalability. We will show how homomorphisms can be efficiently implemented in the MapReduce model, and demonstrate that many existing work on homomorphic approach to the development of parallel programs can be fully utilized in our framework for solving various kinds of problems.
