Concurrency Control, - Presentation - IT & Software PDF Download

CONCURRENCY CONTOL

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

DEFINITION

Concurrency means allowing more than one transaction to run simultaneously on the same database.
When several transactions run concurrently database consistency can be destroyed.
To remove this problem Concurrency control is used.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Need of Concurrency Control (CC)

The concurrent execution of transactions may lead, if uncontrolled, to problems such as an inconsistent database.
CC techniques are used to ensure that multiple transactions submitted by various users do not interfere with one another in a way that produces incorrect results.
The effect on a database of any number of transactions executing in parallel must be the same as if they were executed one after another.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Advantages of Concurrent Execution of Transactions

Improvement Throughput
Reduced Waiting Time

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Problems of Concurrent Execution of Transactions

For Concurrency problems following control mechanism is required:
Lost Update Problem
The Temporary Update (Uncommitted Dependency) Problem
The Incorrect Summary (Inconsistent Analysis) Problem

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

The Lost Update Problem

In this problem, 2 transactions accessing the same database item have their operations interleaved in a way that makes the database
item incorrect.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Consider the following Example

Lost Update Problem…

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

The Temporary Update (Uncommitted Dependency) Problem

This problem occurs when one transaction is allowed to see the intermediate results of another transaction before it is committed.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Consider the Example - The Temporary Update (Uncommitted Dependency) Problem

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

The Incorrect Summary (Inconsistent Analysis) Problem

This problem occurs when a transaction reads several values from a database while a second transaction updates some of them.
For e.g. Values of variable A, B, C and Sum are in column 3,4,5 and 6. The initial values of Sum is 100, 50, 25 and 0 respectively.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

The Incorrect Summary (Inconsistent Analysis) Problem

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Concurrency Control Schemes

Concurrency control schemes are divided into 2 categories:
Pessimistic or Conservative Approach
Optimistic Approach

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Pessimistic Approach

This approach says that there must be some concurrency control techniques deployed before the transactions are allowed to access the database.
Methods of Pessimistic approach:
Locking Protocol
Time Stamp Based Protocol

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Locking for Concurrency Control

Locking
It is a procedure used to control concurrent access to data. In this method when one transaction is accessing the database, a
Lock may deny access to other transactions to produce incorrect results.
Lock
It is a variable associated with a data item. It describes the status of the item with respect to possible operation that can be applied to it.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Types of Lock

Binary Lock-Two States (Lock and Unlock)
Share/Exclusive Lock (Read/Write)

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Locking Operations

Read_lock(A)= Lock-S(A)
Write_lock(A)=Lock-X(A)
Unlock(A)
S-Shared Lock
X-Exclusive Lock

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Compatibility of Locks

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Locking Example

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Problems with Locking

Dead Lock:
It happens whenever a transaction waits for a lock to be unlock (to access the data).
Problem of Starvation
When the data requested by 1 transaction is held by some other transactions again and again and the requested data is not given.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Starvation Example

T2 holds data item on Shared-mode lock.
T1 request Exclusive-mode lock on same data item.
T1 has to wait while T2 release it.
Meanwhile T3 requests same data item for Shared- mode lock and gets it from T2.
T1 still waiting.
Now T4 requests same data item for Shared-mode lock and gets it from T3.
T1 still waiting and is said to be Starved.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Pessimistic Execution

Validate
Read
Compute
Write

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Problems in Two-Phase Locking

Deadlock
Cascading roll back

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Time-Stamp (TS) based Protocol

In this, a unique fixed timestamp is associated with each transaction to keep the order of the transaction. It is denoted by TS (T1).

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Example

If a transaction T1 has been assigned timestamp TS (T1) and a new transaction TS (T2) enters the system , then TS (T1) < TS (T2)

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Two methods for implementing TS

Use the value of the system as the timestamp (System Clock).
Use a logical counter that is incremented after a new timestamp has been assigned.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Implementation Method

W-timestamp (Q) denotes the largest TS of any transaction that executed write (Q) successfully.
R-timestamp (Q) denotes the largest TS of any transaction that executed read (Q) successfully.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Optimistic Approach

It allows transaction to proceed in unsynchronized way and only checks/locks conflicts at the end.
Based on idea that conflicts are rare.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Validation Based Protocol

It consist of 3 phases depending upon whether it is Read only transaction or Read-Write transaction.
Phases are:
Read Phase
Validation Phase
Write Phase

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Validation Based Protocol Phases

Read Phase:
In this every transaction reads the values of all the data elements it needs from the database and stores them in Local variables.
All updates are applied to the Local copy of the data and not to the original database.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Validation Based Protocol Phases…

Validation Phase:
Come after end of Read phase.
Certain checks are performed to ensure that no conflict has occurred.
For Read only transactions this phase consist of checking that the data elements read are ok, no conflict is there then it is Committed.
If conflict is there then transaction is Aborted or Restarted.
For Updating transactions, it checks whether current transaction leaves database in a consistent state, if not then transaction is Aborted.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Validation Based Protocol Phases…

Write Phase:
This phase is for only Read-Write transaction not for Read only transaction.
If the transaction has passed the validation phase successfully then all the changes made by the transaction to the Local copy are made to Final database.

Concurrency Control ------------------------------------------------------------------------------------- Next Slide

Optimistic Execution

Read
Compute
Validate
Write