Browse Prior Art Database

Data Base Integrity at Emergency Restart in Data Sharing

IP.com Disclosure Number: IPCOM000046517D
Original Publication Date: 1983-Jul-01
Included in the Prior Art Database: 2005-Feb-07
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Yamashita, A: AUTHOR

Abstract

In this article I describe a method for operating a plurality of digital computers sharing access to a common data base according to a locking protocol which enables, during emergency restart, the identification of locks owned by each computer at the time of a prior failure. Most recently used buffers are located by examining log (or journal) data, and then whether or not committed data in the buffers had been written to a data base control interval (or block) on disk storage is determined by comparing a control interval update sequence number (CUSN) in the data base block with that in the log. If the CUSN in the data base block is one less than the CUSN in the journal, then a lock was held at the time of failure.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 57% of the total text.

Page 1 of 1

Data Base Integrity at Emergency Restart in Data Sharing

In this article I describe a method for operating a plurality of digital computers sharing access to a common data base according to a locking protocol which enables, during emergency restart, the identification of locks owned by each computer at the time of a prior failure. Most recently used buffers are located by examining log (or journal) data, and then whether or not committed data in the buffers had been written to a data base control interval (or block) on disk storage is determined by comparing a control interval update sequence number (CUSN) in the data base block with that in the log. If the CUSN in the data base block is one less than the CUSN in the journal, then a lock was held at the time of failure.

In a data base management system without data sharing, a restarting system can UNDO/REDO all the data that are potentially not externalized to a data base control interval on non-volatile storage as long as that restarting system owns the data base. In a data sharing environment, UNDO/REDO can only be done to those data that are still owned (locked) at the time of the failure. While such locks are certainly held as long as data changes to the data base are not externalized to the physical media (that is, non-volatile storage), information on locks may be lost as a result of a failure if, for example, the lock manager fails. By the method herein described, locks owned at the time of the failure are...