What is Read-Write conflict in the database? Why does it occur in the DBMS (Database Management System)?
Data is crucial for any kind of project. Text data required for the project can be saved in a file or in any reliable database. Days are gone when we were using text files system to save the data.
In a database management system (DBMS), data is saved in the database. There are many advantages of using a DBMS system over the file system.
In this article, we will see all the different types of read write conflict in DBMS.
Table of Contents
Data is saved into the database. And to fetch and write data from the database, you need to perform read and write operation on the data.
There are multiple entities (multiple users, multiple APPs, multiple websites…) performing these multiple operations on the same data at the same time.
Just as an instance, there is a centralized banking database where all the information about the users and the bank related information is saved.
There might be multiple user logins to their bank account and access the same bank information at the same time. They perform various operations like reading or updating or deleting the data.
So basically there are two types of operations that can be performed on the database.
Read the data value from the database. It is a safe operation, as it does not update any information in the database.
It writes data into the database and saved it for further use. After writing data, data has to be committed (Commit Operation) to making updated information available for further operations. This operation is more prone to vulnerability as it involves modifying database information.
For a data update operation, it involves both the reading and writing operation.
The set of these multiple operations/instruction (to accomplish a particular task) are performed on the data. This set is called as Transaction.
Every entity accessing data holds different transaction and all the transactions execute simultaneously.
The operations involved in the successful Transaction:
While performing these operations by the different users on the same data at the same time may arise data conflict. As like, accessing data by one transaction before committing another transaction does not hold good and gives data conflict.
There are different types of data conflict happens during reading and writing operations called as “Read-Write conflict”.
As I mentioned it earlier, the read operation is safe as it does modify any information. So, there is no Read-Read (RR) conflict in the database.
Any number of transactions are free to read the same data (without conflict) anytime as long as there is no write operation.
So, there are three types of conflict in the database transaction.
Let’s see one-by-one with the example.
This conflict occurs when a transaction read the data which is written by the other transaction before committing.
In the following diagram, I am using the following notions
A, B - two different data objects from database T1, T2 - two different transactions R(A) - reading data A W(A) - Writing data A Com. - committing transaction
Here, the transaction T2 is reading the data which is written by the T1 before T2 commits. It is also called as Dirty Read.
It violates the ACID property of data consistency rule.
The new transaction (says T2) should not read the value if any other another transaction (say T1) already has written the data and has not committed.
Transaction T2 is Writing data which is previously read by transaction T1.
Here if you look at the diagram above, data read by transaction T1 before and after T2 commits is different.
Suppose Alice and Bob want to book the flight for their vacation. Alice open airlines website to check the availability and cost of the ticket. There is only one ticket is available. Alice finds it expensive and looks for other airlines fares if she gets any offers.
Meanwhile, Bob logins to the same airline portal and book the ticket. Now, there is no more flight ticket available.
Alice does not find any good offer on other airlines portal, she comes back to the previous airline portal. And tried to book a flight ticket even it is not available. Its Conflict, Read-Write (WR) conflict.
Here Transaction T2 is writing data which is already written by other transaction T1. T2 overwrites the data written by T1. It is also called as a blind write operation.
Data written by T1 has vanished. So it is data update loss.
Alice and Bob share common Google-sheet online. Both read the file. Alice updates the document and forgets to save the file. On another end, Bob updates the Google sheet and save the file.
The content updated by Alice is overwritten by Bob. The data updated by Alice is lost. It is called as Write-Write (WW) conflict.
This is all about different types of read write conflict in DBMS. If you have any doubt or any thought about this article, write a comment below. I will reply you back as soon as possible.