Large applications executing on Grid or cluster architectures consisting of hundreds or thousands of computational nodes create problems with respect to reliability. The source of the problems is node failures and the need for dynamic configuration over extensive runtime.
By allowing recovery even under different numbers of processors, the approaches are especially suitable for applications with a need for adaptive or reactionary configuration control. The low-cost protocols offer the capability of controlling or bounding the overhead. A formal cost model is presented, followed by an experimental evaluation. It is shown that the overhead of the protocol is very small, and the maximum work lost by a crashed process is small and bounded.
Communication Induced Check-pointing protocols usually make the assumption that any process can be check-pointed at any time. An alternative approach which releases the constraint of always check-pointable processes, without delaying any do not message reception nor did altering message ordering enforce by the communication layer or by the application. This protocol has been implemented within Pro-Active, an open source Java middleware for asynchronous and distributed objects implementing the ASP (Asynchronous Sequential Processes) model.
This paper presents two fault-tolerance mechanisms called Theft-Induced Check pointing and Systematic Event Logging. These are transparent protocols capable of overcoming problems associated with both benign faults, i.e., crash faults, and node or subnet volatility. Specifically, the protocols base the state of the execution on a dataflow graph, allowing for efficient recovery in dynamic heterogeneous systems as well as multithreaded applications.
- Networking Module
- Logging Module
- Check-pointing Module
- Work Stealing Module
- Fault and Fault Free Module