An Essential Design Pattern for Fault-Tolerant Distributed State Sharing.

Saved in:

Bibliographic Details
Title:	An Essential Design Pattern for Fault-Tolerant Distributed State Sharing.
Authors:	Islam, Nayeem¹ nayeem@watson.Ibm.com., Devarakonda, Murthy² mdev@watson.ibm.com
Source:	Communications of the ACM. Oct96, Vol. 39 Issue 10, p65-74. 10p. 6 Graphs.
Subjects:	Distributed operating systems (Computers), Fault-tolerant computing, Software patterns, Object-oriented methods (Computer science), Computer software development, Systems design
Abstract:	The article focuses on the design and implementation of distributed computer systems. Despite significant common experience and a good deal of success, the cost and complexity of distributed programming continues to be high. Increasingly, project managers look to object-oriented (OO) programming to improve productivity in this area. In order for distributed programming to derive benefits from OO, encouraging the reuse of proven distributed solutions is crucial. In developing distributed programs, successful designers have focused on two important requirements: performance and fault tolerance. Therefore, distributed design patterns should highlight the aspects of a solution that address these requirements. One such OO design pattern, the Recoverable Distributor, allows mixing and matching of fault-tolerance and data-consistency protocols and has been used in a distributed scheduler and a distributed lock manager. Recoverable Distributor is an essential design pattern in fault-tolerant, state-sharing distributed programs. It emphasizes the significance of performance, fault detection, and recovery in such programs. This article presents this pattern and describes how it is used in two commercial distributed programs.
Database:	Engineering Source

Description
Abstract:	The article focuses on the design and implementation of distributed computer systems. Despite significant common experience and a good deal of success, the cost and complexity of distributed programming continues to be high. Increasingly, project managers look to object-oriented (OO) programming to improve productivity in this area. In order for distributed programming to derive benefits from OO, encouraging the reuse of proven distributed solutions is crucial. In developing distributed programs, successful designers have focused on two important requirements: performance and fault tolerance. Therefore, distributed design patterns should highlight the aspects of a solution that address these requirements. One such OO design pattern, the Recoverable Distributor, allows mixing and matching of fault-tolerance and data-consistency protocols and has been used in a distributed scheduler and a distributed lock manager. Recoverable Distributor is an essential design pattern in fault-tolerant, state-sharing distributed programs. It emphasizes the significance of performance, fault detection, and recovery in such programs. This article presents this pattern and describes how it is used in two commercial distributed programs.
ISSN:	00010782
DOI:	10.1145/236156.236172