1. Chapter 15 Transaction Management

2. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Outline Transaction basics Concurrency control Recovery management Transaction design issues Workflow management

3. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Definition Supports daily operations of an organization Collection of database operations Reliably and efficiently processed as one unit of work No lost data –Interference among multiple users –Failures

4. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Airline Transaction Example START TRANSACTION Display greeting Get reservation preferences from user SELECT departure and return flight records If reservation is acceptable then UPDATE seats remaining of departure flight record UPDATE seats remaining of return flight record INSERT reservation record Print ticket if requested End If On Error: ROLLBACK COMMIT

5. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Properties Atomic: all or nothing Consistent: database must be consistent before and after a transaction Isolated: no unwanted interference from other users Durable: database changes are permanent after the transaction completes

6. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Processing Services Concurrency control Recovery management Service characteristics –Transparent –Consume significant resources –Significant cost component

7. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Concurrency Control Problem definition Concurrency control problems Concurrency control tools

8. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Concurrency Control Problem Objective: –Maximize work performed –Throughput: number of transactions processed per unit time Constraint: –No interference: serial effect –Interference occurs on commonly manipulated data known as hot spots

9. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Lost Update Problem

10. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Uncommitted Dependency Problem

11. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Inconsistent Retrieval Problems Interference causes inconsistency among multiple retrievals of a subset of data Incorrect summary Phantom read Non repeatable read

12. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Incorrect Summary Problem

13. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Locking Fundamentals Fundamental tool of concurrency control Obtain lock before accessing an item Wait if a conflicting lock is held –Shared lock: conflicts with exclusive locks –Exclusive lock: conflicts with all other kinds of locks Concurrency control manager maintains the lock table

14. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Locking Granularity

15. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Deadlock (Mutual Waiting)

16. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Deadlock Resolution Detection –Can involve significant overhead –Not widely used Timeout –Waiting limit –Can abort transactions that are not deadlocked –Widely used although timeout interval is difficult to determine

17. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Two Phase Locking (2PL) Protocol to prevent lost update problems All transactions must follow Conditions –Obtain lock before accessing item –Wait if a conflicting lock is held –Cannot obtain new locks after releasing locks

18. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. 2PL Implementation

19. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Optimistic Approaches Assumes conflicts are rare No locks Check for conflicts –After each read and write –At end of transaction Evaluation –Less overhead –More variability

20. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Recovery Management Device characteristics and failure types Recovery tools Recovery processes

21. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Storage Device Basics Volatile: loses state after a shutdown Nonvolatile: retains state after a shutdown Nonvolatile is more reliable than volatile but failures can cause loss of data Use multiple levels and redundant levels of nonvolatile storage for valuable data

22. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Failure Types Local –Detected and abnormal termination –Limited to a single transaction Operating System –Affects all active transactions –Less common than local failures Device –Affects all active and past transactions –Least common

23. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Log History of database changes Large storage overhead Operations –Undo: revert to previous state –Redo: reestablish a new state Fundamental tool of recovery management

24. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Log Example

25. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Checkpoints Reduces restart work but adds overhead –Checkpoint log record –Write log buffers and database buffers Checkpoint interval: time between checkpoints Types of checkpoints –Cache consistent –Fuzzy

26. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Other Recovery Tools Force writing –Checkpoint time –End of transaction Database backup –Complete –Incremental

27. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Recovery from a Media Failure Restore database from the most recent backup Redo all committed transactions since the most recent backup Restart active transactions

28. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Recovery Timeline

29. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Recovery Processes Depend on timing of database writes Immediate update approach: –Before commit –Log records written first (write-ahead log protocol) Deferred update approach –After commit –Undo operations not needed

30. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Immediate Update Recovery

31. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Deferred Update Recovery

32. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Design Issues Transaction boundary Isolation levels Deferred constraint checking Savepoints

33. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Boundary Decisions Division of work into transactions Objective: minimize transaction duration Constraint: enforcement of important integrity constraints Transaction boundary decision can affect hot spots

34. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Registration Form Example

35. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Transaction Boundary Choices One transaction for the entire form One transaction for the main form and one transaction for all subform records One transaction for the main form and separate transactions for each subform record

36. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Isolation Levels Degree to which a transaction is separated from the actions of other transactions Balance concurrency control overhead with interference problems Some transactions can tolerate uncommitted dependency and inconsistent retrieval problems Specify using the SET TRANSACTION statement

37. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. SQL Isolation Levels LevelXLocksSLocksPLocksInterference Read uncommitted None Uncommitted dependency Read committed LongShortNoneAll except uncommitted dependency Repeatable read Long Short (S), Long (X) Phantom reads SerializableLong None

38. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Scholar’s Lost Update Transaction ATimeTransaction B Obtain S lock on SRT1T1 Read SR (10)T2T2 Release S lock on SRT3T3 If SR > 0 then SR = SR -1T4T4 T5T5 Obtain S lock on SR T6T6 Read SR (10) T7T7 Release S lock on SR T8T8 If SR > 0 then SR = SR -1 Obtain X lock on SRT9T9 Write SR (9)T 10 CommitT 11 T 12 Obtain X lock on SR T 13 Write SR (9)

39. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Integrity Constraint Timing Most constraints checked immediately Can defer constraint checking to EOT SQL SET CONSTRAINTS statement

40. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Save Points Some transactions have tentative actions SAVEPOINT statement determines intermediate points ROLLBACK to specified save points

41. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Workflow Management Workflow description Enabling technologies Advanced transaction management

42. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Workflows Set of tasks to accomplish a business process Human-oriented vs. computer-oriented –Amount of judgment –Amount of automation Task structure vs. task complexity –Relationships among tasks –Difficulty of performing individual tasks

43. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Enabling Technologies Distributed object management –Many kinds of non traditional data –Data often dispersed in location Workflow modeling –Specification –Simulation –Optimization

44. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Advanced Transaction Management Conversational transactions Transactions with complex structure Transactions involving legacy systems Compensating transactions More flexible transaction processing

45. McGraw-Hill/Irwin © 2004 The McGraw-Hill Companies, Inc. All rights reserved. Summary Transaction: user-defined collection of work DBMSs support ACID properties Knowledge of concurrency control and recovery important for managing databases Transaction design issues are important Transaction processing is an important part of workflow management