《数据库系统原理Database System Concepts》双语课件-ch11(concurrency control).pptVIP

2007-6-16 APWeb/WAIM2007 Huang Shan 2007-6-16 APWeb/WAIM2007 Huang Shan 2007-6-16 Chapter Outline Possible Problems of Concurrency Lock_based Methods Two-phase Locking Protocol Deadlock Timestamp Methods Validation-Based Protocol Possible Problems of concurrency Lost update problem Reading uncommitted data ( “dirty reads”) problem Incorrect summary problem Example transactions T1: Transfer $50 from account A to B Read (A) A := A – 50 Write (A) Read(B) B := B + 50 Write(B) Problem I: Lost update Why Concurrency Control is needed? The Lost Update Problem This occurs when two transactions that access the same database items have update operations interleaved in a way that makes the value of some database item incorrect. The Dirty Read Problem This occurs when one transaction updates a database item and then the transaction fails for some reason. The updated item is accessed by another transaction before it is changed back to its original value. The Incorrect Summary Problem If one transaction is calculating an aggregate summary function on a number of records while other transactions are updating some of these records, the aggregate function may calculate some values before they are updated and others after they are updated. Which kind of problem? Discussion What are yours solutions to solve the problems? ——Lost Update ——Dirty Read ——Incorrect Summary Solutions for Concurrency Control Pessimistic methods Lock-based protocols Timestamp-based protocols Optimistic methods Validation-Based Protocols Lock-based Methods The lock_based methods use Lock(X) before accessing data object X, Unlock(X) after accessing the object: Lock (X): Locking is an operation which request permissions to access X. Unlock (X): Unlocking is an operation which removes these permissions from X. Lock and Unlock are atomic operations. Basic Lock Model Two locks modes: shared (S) (read) exclusive (X) (write) S mode More than one transaction can apply share lock on A