DocumentCode :
2175486
Title :
Efficient Incremental Optimal Chain Partition of Distributed Program Traces
Author :
Ikiz, Selma ; Garg, Vijay K.
Author_Institution :
University of Texas at Austin
fYear :
2006
fDate :
2006
Firstpage :
18
Lastpage :
18
Abstract :
An important problem in distributed systems is observation of global properties of distributed computations. What makes this problem difficult is that events in the computation can be concurrent, i.e. the relation between events forms a partial order, not a total order. One of the fundamental parameters of a partial order is the width, which corresponds to the maximum number of mutually incomparable elements. For example, a process-time diagram that shows this partial order decomposition in minimum number of chains can be very useful in monitoring or debugging such computations. In this paper, we present an incremental algorithm to compute the optimal chain partition. We compare our algorithm with existing chain reduction algorithms. From a practical point of view, performance evaluation shows that our approach achieves up to 90% run-time improvement over the previously known algorithms.
Keywords :
Clocks; Computer science; Concurrent computing; Debugging; Distributed computing; Monitoring; Partitioning algorithms; Runtime; Testing; Visualization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Distributed Computing Systems, 2006. ICDCS 2006. 26th IEEE International Conference on
ISSN :
1063-6927
Print_ISBN :
0-7695-2540-7
Type :
conf
DOI :
10.1109/ICDCS.2006.34
Filename :
1648805
Link To Document :
بازگشت