K锁算法——一种数据库上的可适应并发控制策略  

K-LOCKING ALGORITHM: AN ADAPTIVE CONCURRENCY CONTROL STRATEGY IN DATABASE

作　　者：周建强[1] 谢立[1] 孙钟秀[1] 朱根江[1] 

机构地区：[1]南京大学计算机科学系,南京210008

出　　处：《计算机学报》1991年第12期933-941,共9页Chinese Journal of Computers

摘　　要：本文提出了一种基于悲观和乐观思想的并发控制算法——K锁算法.K锁算法利用K锁机制,在一定程度上允许冲突执程的并发执行.和乐观算法比较,它减少了在系统的执程冲突概率较大时执程的重执次数,和悲观算法比较,它提高了冲突执程的执行并发度.There are two basic approaches, optimistic and pessimistic, for ensuringthe consistency of databases. The former performs better when the degree of transaction interference is low while the reverse holds. A new strategy called K-lock algorithm is prtsented. It is based on the two-phase lock method and an optimistic approach, and allows conflict transactions to execute concurrently in a limited extent. The K-lock algorithm achieves increase in degree of concurrency over the two-phase lock method, and has less rollbacks of invalid transactions than the optimistic approach in high degree of transaction interference.

关 键 词：数据库 并发控制 K锁算法 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]