Deadlocks System Model Deadlock Characterization Methods For Handling Deadlocks Deadlock Prevention Deadlock Avoidance Deadlock Detection And Recovery From Deadlock Pdf

Deadlock Probability Prediction and Detection in Distributed Systems

The problem of deadlocks in distributed systems can be handled by prevention, avoidance and detection strategies. All methods require modeling the system by Resource Allocation graphs or Transaction-Wait-For graphs. The current methods can fail to detect deadlocks in some system scenarios and also do not provide numerical values for system deadlock probability. Analysis of complex distributed systems with varying number of processes and resources requires a computerized solution. We developed a General Transaction program, based on a technique defined as Resource Allocation Arrays, which provides a framework for representing process resource interactions in the system. The State Identifier program uses this framework to develop system state space diagrams. The State Identifier program also identifies the deadlock states in the system. Models developed in OPNET simulate the system state space and provide numerical values for probability of being in any state of the system, and in effect the probability of reaching a system deadlock. Data shows that the simulation models are 99.86% accurate. The developed computerized approach substantially reduces the numerical analysis and computational time for analyzing such complex system scenarios.

On-line Deadlock Detection in Distributed Computer Systems

A new algorithm, the Horizontal and Vertical Algorithm, for on-line detection of deadlocks in distributed computer systems, is presented. Two protocols for implementing the algorithm are given. The first protocol, the centralized protocol, is based on the assumption that one site in the network acts as the controller for global resource allocation and deadlock detection. The second protocol, the distributed protocol, distributes the responsibilities of resource allocation and deadlock detection among the sites where the requested resources reside. The new deadlock detection protocols have two important features. Both protocols are characterized by their simplicity in implementation as compared to most published protocols. The storage requirement needed to run the distributed protocol is considerably reduced. The distributed protocol is also characterized by a significant reduction of communication messages passed around the different sites in the network. The new algorithm is compared with the distributed algorithm proposed by Barry Goldman and the preemption method of deadlock prevention on a ring network. The comparison was made by means of simulation models. Simulation models are developed for both the centralized and distributed control of the new algorithm, Goldman's algorithm and the preemption technique. The performances of the algorithms are measured in terms of process response tim--average delay per process, and process throughput--the number of processes completed per unit time. Resource request response time--average time to process a resource request and throughput--the number of requests processed per unit time are also measured. Communication overhead associated with the use of each algorithm and frequency of deadlock occurrence are also measured. The simulation results, for the distributed Horizontal and Vertical algorithm, are used to develop an M/M/z queueing model to measure the request response time of the algorithm. This is done by a regression technique. The results of the analytical model show a very close fit with the results of the simulation model.

Development of a Hybrid "avoidance-deadlock Detection and Recovery" Scheme for Distributed Logic Simulation