Abstract—Distributed nature of transactions arising at different sites and needing resources from diverse locations pose various operational problems, such as deadlocks, concurrency and data recovery. A deadlock may occur when a transaction enters into wait state that request resource from other blocked transactions. Deadlock detection and resolving is very difficult in a distributed database system because no controller has complete and current information about the system and data dependencies. In this paper, an enhanced technique for deadlock resolution is presented, which minimizes the abortion or waiting of the selected victim transactions. The proposed system includes the use of fuzzy logic by creating a set of fuzzy rules in order to deal with criticalness and similarity attributes of transactions. By using these rules, fuzzy logic will try to provide an easy conflict resolution method between transactions to diminish transactions wasted restart, and guaranteeing temporal consistency of data and transactions. Furthermore, the presented deadlock handling algorithm does not detect any false deadlock or exclude any really existing deadlocks. Experimental results show performance of the recommended system benefits such as increase in commit rate and decrease in re-execution or waiting of the transactions.
Index Terms—Fuzzy logic, deadlock resolving, transactions conflict, concurrency control.
Saad M. Darwish and Adel A. El-Zoghabi are with the Department of Information Technology, Institute of Graduate Studies and Research, Alexandria University, 163 Horreya Avenue, El-Shatby 21526, P.O. Box 832, Alexandria, Egypt (e-mail: saad.saad@alexu.edu.eg, zoghabi@gmail.com).
Marwan H. Hassan is with the College of Imam Aladham, Branch Anbar province, The Sunni Endowment Diwan, Iraq (e-mail: marwanhh80@yahoo.com).
[PDF]
Cite: Saad M. Darwish, Adel A. El-Zoghabi, and Marwan H. Hassan, "Soft Computing for Database Deadlock Resolution," International Journal of Modeling and Optimization vol. 5, no. 1, pp. 15-21, 2015.
