Failure detection and membership management in grid environments

Failure detectors are an integral part of any fault tolerant distributed system and hence have been a well-studied area. However, earlier proposed failure detectors fail to perform efficiently when applied to grid environments. Most of the earlier proposed detectors were either designed for local area networks or to handle small number of nodes and hence lack in areas such as scalability, efficiency, running times etc. In this paper we propose a highly scalable failure detector protocol that is aided by a membership management service. The membership management service is essential to make the failure detector transparent to changes in the system. Using a distributed heartbeat mechanism, for an unreliable failure detector, we have overcome the shortcomings of similar schemes proposed earlier. It realizes scalability by reducing context switching requirements and achieves faster failure detection. The membership management protocol handles membership issues with a worst case complexity of O(n) where n is the number of heartbeat groups. Note that n is much smaller than the total number of nodes in the grid. The algorithm is also shown to be failure resilient and scalable.