Chapter Three

Transaction Management and Concurrency Control

What is a Transaction?
 Any action that reads from and/or writes to a database may consist of:
o Simple SELECT statement to generate list of table contents
o Series of related UPDATE statements to change values of attributes in various tables
o Series of INSERT statements to add rows to one or more tables
 Combination of SELECT, UPDATE, and INSERT statements
 Transaction is logical unit of work that must be either entirely completed or aborted
 Successful transaction changes database from one consistent state to another
o One in which all data integrity constraints are satisfied
 Most real-world database transactions are formed by two or more database requests
o Equivalent of a single SQL statement in an application program or transaction

Evaluating Transaction Results

 Not all transactions update database
 SQL code represents a transaction because database was accessed
 Improper or incomplete transactions can have devastating effect on database integrity
o Some DBMSs provide means by which user can define enforceable constraints
o Other integrity rules are enforced automatically by the DBMS
Transaction Properties
 Atomicity
o Requires that all operations (SQL requests) of a transaction be completed
 Consistency
o Indicates the permanence of database’s consistent state
 Isolation
o Data used during execution of a transaction cannot be used by second transaction until
first one is completed
 Durability
o Indicates permanence of database’s consistent state Isolation
 Serializability
o Ensures that concurrent execution of several transactions yields consistent results
Transaction Management with SQL
 ANSI has defined standards that govern SQL database transactions
 Transaction support is provided by two SQL statements: COMMIT and ROLLBACK
 ANSI standards require that, when a transaction sequence is initiated by a user or an
application program, it must continue through all succeeding SQL statements until one of
four events occurs
 COMMIT statement is reached
 ROLLBACK statement is reached
 End of program is reached
 Program is abnormally terminated
The Transaction Log
 Transaction log stores:
 A record for the beginning of transaction
 For each transaction component (SQL statement):
 Type of operation being performed (update, delete, insert)
 Names of objects affected by transaction
 “Before” and “after” values for updated fields
 Pointers to previous and next transaction log entries for the same transaction
  Ending (COMMIT) of the transaction
Concurrency Control
• Coordination of simultaneous transaction execution in a multiprocessing database system
• Objective is to ensure serializability of transactions in a multiuser database environment
• Simultaneous execution of transactions over a shared database can create several data integrity
and consistency problems
• Lost updates
• Uncommitted data
• Inconsistent retrievals
Lost Updates
 Normal Execution of Two Transactions
Uncommitted Data
 Correct Execution of two Transactions
 An Uncommitted Data Problem
Inconsistent Retrievals
 Retrieved During Update
 Transaction Results: Data entry Correction
The Scheduler
 Special DBMS program
 Purpose is to establish order of operations within which concurrent transactions are
executed
 Interleaves execution of database operations to ensure serializability and isolation of
transactions
 Bases its actions on concurrency control algorithms
 Ensures computer’s central processing unit (CPU) is used efficiently
 Facilitates data isolation to ensure that two transactions do not update same data element at
same time
 Read/Write conflict Scenarios: Conflicting DB Operations matrix
Concurrency Control with Locking Methods
 Lock
 Guarantees exclusive use of a data item to a current transaction
 Required to prevent another transaction from reading inconsistent data
 Lock manager
 Responsible for assigning and policing the locks used by transactions
Lock Granularity
 Indicates level of lock use
 Locking can take place at following levels:
 Database  Row
 Table  Field (attribute)
 Page
 Database-level lock
 Entire database is locked
 Table-level lock
 Entire table is locked
 Page-level lock
 Entire disk page is locked
 Row-level lock
 Allows concurrent transactions to access different rows of same table, even if rows are
located on same page
 Field-level lock
  Allows concurrent transactions to access same row, as long as they require use of
different fields (attributes) within that row
 Database-level locking sequence
 An example of a table level lock
 An example of a page level lock
 An example of a row level lock
Lock Types
 Binary lock
 Has only two states: locked (1) or unlocked (0)
 Exclusive lock
 Access is specifically reserved for transaction that locked object
 Must be used when potential for conflict exists
 Shared lock
 Concurrent transactions are granted Read access on basis of a common lock
 An example of a binary lock
Two-Phase Locking to Ensure Serializability
 Defines how transactions acquire and relinquish locks
 Guarantees serializability, but it does not prevent deadlocks
 Growing phase - Transaction acquires all required locks without unlocking any data
 Shrinking phase - Transaction releases all locks and cannot obtain any new lock
 Governed by the following rules:
 Two transactions cannot have conflicting locks
 No unlock operation can precede a lock operation in the same transaction
 No data are affected until all locks are obtained—that is, until transaction is in its locked
point
Deadlocks
 Condition that occurs when two transactions wait for each other to unlock data
 Possible only if one of the transactions wants to obtain an exclusive lock on a data item
 No deadlock condition can exist among shared locks
 Control through:
 Prevention
 Detection
 Avoidance
Concurrency Control with Time Stamping Methods
 Assigns global unique time stamp to each transaction
 Produces explicit order in which transactions are submitted to DBMS
 Uniqueness
 Ensures that no equal time stamp values can exist
 Monotonicity
 Ensures that time stamp values always increase
Wait/Die and Wound/Wait Schemes
 Wait/die
 Older transaction waits when requests lock first.
 Younger is rolled back when requests lock first.
 Wound/wait
 Older transaction rolls back younger transaction when requests lock first.
 Younger transaction waits when requests lock first.
Concurrency Control with Optimistic Methods
 Optimistic approach
 Based on assumption that majority of database operations do not conflict
 Does not require locking or time stamping techniques
  Transaction is executed without restrictions until it is committed
 Phases are read, validation, and write
Database Recovery Management
 Database recovery
o Restores database from given state, usually inconsistent, to previously consistent state
o Based on atomic transaction property
 All portions of transaction must be treated as single logical unit of work, so all
operations must be applied and completed to produce consistent database
o If transaction operation cannot be completed, transaction must be aborted, and any
changes to database must be rolled back (undone)
Transaction Recovery
 Makes use of deferred-write and write-through techniques
 Deferred write
 Transaction operations do not immediately update physical database
 Only transaction log is updated
 Database is physically updated only after transaction reaches its commit point using
transaction log information
 Write-through
 Database is immediately updated by transaction operations during transaction’s
execution, even before transaction reaches its commit point
Summary
 Transaction
 Sequence of database operations that access database
 Represents real-world events
 Must be logical unit of work
 No portion of transaction can exist by itself
 Takes database from one consistent state to another
 One in which all data integrity constraints are satisfied
 Transactions have five main properties: atomicity, consistency, isolation, durability, and
serializability .
 SQL provides support for transactions through the use of two statements: COMMIT and ROLLBACK
 SQL transactions are formed by several SQL statements or database requests
 ransaction log keeps track of all transactions that modify database
 Concurrency control coordinates simultaneous execution of transactions
 Scheduler is responsible for establishing order in which concurrent transaction operations are
executed
 Lock guarantees unique access to a data item by transaction
 Two types of locks can be used in database systems: binary locks and shared/exclusive locks
 Serializability of schedules is guaranteed through the use of two-phase locking
 When two or more transactions wait indefinitely for each other to release lock, they are in
deadlock, or deadly embrace
 Three deadlock control techniques: prevention, detection, and avoidance
 Concurrency control with time stamping methods assigns unique time stamp to each transaction
and schedules execution of conflicting transactions in time stamp order
 Concurrency control with optimistic methods assumes that the majority of database transactions
do not conflict and that transactions are executed concurrently, using private copies of the data
 Database recovery restores database from given state to previous consistent state