On detecting over-eager concurrency in asynchronously communicating concurrent object systems

In this paper we show that systems developed using active objects and asynchronous method calls can result in system failure due to over-eager concurrency, which we call flooding. A system may feed an object with more calls than it is able to handle, in some cases even regardless of its processing speed. We refer to this situation as flooding of the object. We distinguish between strong and weak flooding. In particular, the notion of strong flooding could lead to problems such as non-responsive objects, system crash, overfull buffers or massive amounts of lost messages, even in the presence of fair scheduling. We present an algorithm to statically detect strong and weak flooding, and prove the soundness of the algorithm.