Abstract

"From a certain point onward there is no longer any turning back. That is the point that must be reached." Kafka Protocols that allow operational sites to continue transac-tion processing even though site failures have occurred are called nonblocking. Many applications require nonblocking Qrotocols. This paper investigates the properties of non-blocking protocols. Necessary and sufficient conditions for a protocol to be nonblocking are presented and from these conditions a method for designing them is derived. Both a central site nonblocking protocol and a decentralized non-blocking protocol are presented.

Citations

266	Notes on database operating systems - Gray - 1978
91	Concurrency control and consistency of multiple copies of data in distributed ingres - Stonebraker - 1979
73	Physical integrity in a large segmented database - Lorie - 1977
24	Reliability Mechanisms for SDD-1: A System for Distributed Databases - Hammer, Shipman - 1980
12	Crash recovery in a distributed storage system - LAMPSON, STURGIS - 1979
11	Notes on distributed databases - Lindsay, Selinger, et al. - 1979
1	Notes on Database Operating Systems," in Systems: An Advanced Springer-VeTlag - Gray - 1979