2002Jeongbo gwahaghoe nonmunji. si'seu'tem mich i'lonRequires access

A Distributed Deadlock Detection and Resolution Algorithm for the OR Model

Soojung Lee

Open publisher page 1 citations

Abstract

Deadlock detection in distributed systems is considered difficult since no single site knows the exact information on the whole system state. This paper proposes a time-efficient algorithm for distributed deadlock detection and resolution. The initiator of the algorithm propagates a deadlock detection message and builds a reduced wait-for graph from the information carried by the replies. The time required for deadlock detection is reduced to half of that of the other algorithms. Moreover, any deadlock reachable from the initiator is detected whereas most previous algorithms only find out whether the initiator is involved in deadlock. This feature accelerates the detection of deadlock. Resolution of the detected deadlock is also simplified and precisely specified, while the current algorithms either present no resolution scheme or simply abort the initiator of the algorithm upon detecting deadlock.

About this research paper

What this paper is about

Deadlock detection in distributed systems is considered difficult since no single site knows the exact information on the whole system state. This paper proposes a time-efficient algorithm for distributed deadlock detection and resolution. The initiator of the algorithm propagates a deadlock detection message and builds a reduced wait-for graph from the information carried by the replies. The time required for deadlock detection is reduced to half of that of the other algorithms. Moreover, any deadlock reachable from the initiator is detected whereas most previous algorithms only find out whether the initiator is involved in deadlock. This feature accelerates the detection of deadlock. Resolution of the detected deadlock is also simplified and precisely specified, while the current algorithms either present no resolution scheme or simply abort the initiator of the algorithm upon detecting deadlock.

Why it matters

OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Deadlock detection in distributed systems is considered difficult since no single site knows the exact information on the whole system state. This paper proposes a time-efficient algorithm for distributed deadlock detection and resolution. The initiator of the algorithm propagates a deadlock detection message and builds a reduced wait-for graph from the information carried by the replies. The time required for deadlock detection is reduced to half of that of the other algorithms. Moreover, any deadlock reachable from the initiator is detected whereas most previous algorithms only find out whether the initiator is involved in deadlock. This feature accelerates the detection of deadlock. Resolution of the detected deadlock is also simplified and precisely specified, while the current algorithms either present no resolution scheme or simply abort the initiator of the algorithm upon detecting deadlock.

Key concepts: Deadlock prevention algorithms, Deadlock, Wait-for graph, Computer science, Algorithm, Resolution (logic), Distributed computing, Graph

Related papers

Back to paper searchBrowse research topicsOriginal source
A Distributed Deadlock Detection and Resolution Algorithm for the OR Model — Research Paper | ScholarLens