Lesson finished
Unlock all 62 lessons of Everything about Database 2.0Full course ₹40,000 · paid
What a DBMS Should Offer: ACID Properties and Databases vs File Systems
A free lesson from Skill-Lync's Everything about Database 2.0 course, on Grow by Skill-Lync · Last updated 16 Nov 2026 · Session: “Module 1 - Characteristics of DBMS, ACID, Databases vs File system”
What will you learn in this free Everything about Database 2.0 lesson?
The instructor explains why a database management system must support concurrent users, using Instagram likes and bank transfers as examples. It defines the four ACID properties, then lists other DBMS characteristics such as security, GUI tools, integrity and reduced redundancy. It closes with the drawbacks of a DBMS and why a file system falls short.
- Atomicity, consistency, isolation and durability explained
- Why a DBMS must handle concurrent access
- Data redundancy and integrity problems of file systems
- Costs and drawbacks of using a DBMS
What comes after lesson 3?
Later in Skill-Lync's Everything about Database 2.0 course (paid): lesson 4, Module 1 - Types of DB, SQL vs NOSQL; lesson 5, Module 1 - CRUD operation example in MySQL, OLTP, OLAP; lesson 6, Module 2 - Database design, ER model, Entity, Attributes. The full course has 62 lessons (19 h 22 min of video); see the course overview or its price and syllabus.
Lesson transcript
Machine transcript of the lesson audio (lightly cleaned). Technical terms may contain errors; the video is the reference.
Now, let's talk a little bit about what characteristics of a database management system there should be. Now, remember that every database has transactions from concurrent user accesses, meaning that your database management system should make sure that your data which is stored as a form of database should be able to be dealt with by multiple users at the same time and still provide all the goodness of consistency in the data. So, it may be that I do not want that my DBMS should be such that it actually does not allow other users to work on the same database when some users are working on it.
So, let's say for example, if there is Instagram, Instagram should be able to allow multiple people to like the same photograph, right? If people have to wait until other users have liked, then they will never be able to like a particular photograph and update the database that they have liked that particular photograph. So, that database should be able to allow transactions from multiple concurrent users at the same time. Now, to do this, we really need four very important properties from a database management system and those are known as asset properties.
Atomicity, isolation, consistency and durability. These asset properties help maintain the data in a healthy state always. Now, what do we mean by healthy state? That first, the data is not corrupted. That, you know, somebody did something with the data and the data got corrupted or the data is inconsistent. Meaning, somebody reads the data of a particular user and let's say the number of followers of the user today looks 3000. But to somebody else looks 3 million. That's pretty inconsistent, right? You do not want this kind of inconsistency in your database.
So, asset properties really help us to make sure that our data is always in a healthy state. Any updates to the data or any deletions from the data are handled so carefully that it always gives a correct user experience to the users when they are dealing with any web application with the database. Remember, that if you want to maintain the consistency of a database before and after a particular transaction from the user, all these four conditions should be met. Now, what is atomicity? Atomicity in brief means all or nothing.
Either do you do that particular transaction or you don't do. For example, a person wants to transfer some money from one user to another. For example, I want to transfer some money, let's say 1000 rupees to another user. What should happen is that first 1000 rupees should be deducted from my bank account and then that same 1000 rupees should be updated to the bank account of the user whom I want to send. It should not be that the 1000 rupees got deducted from my account but it never got credited to the person whom I want to send.
Because then I would be mad at the bank, right? I want either the entire transaction to complete my bank account to be deducted and that person's bank account to be credited or I do not want anything to happen. I'm okay with the bank telling me that my processing failed and that nothing could happen. Then actually they coming up with me that, you know, now your bank account got deducted but the person's data did not get credited because then we don't know what's the status of those 1000 rupees which got deducted from my bank.
I do not know if I'm going to get it back. I do not know if it's going to get credited to the other person in some time. I do not know anything of the sort and I do not want this kind of confusion, this kind of anxiety, right? We do not want this problem in a database. We want the entire transaction to complete or none of it to complete. So either all parts of the transactions are stored or none of them is stored. We do not want to allow any partial transaction to happen.
That is atomicity. What about consistency? Remember that consistency means that before the transaction, the database has to be consistent with whatever has happened. And then after the transaction, it must stay consistent too. For example, in my database, consistency may mean that if I remove a particular student from the institute, then that student's entire data actually gets deleted from the database, is now archived in a historical database. I do not want that student's presence in any of the tables in my database. That could be a consistent thing.
Before the student is present, then I am okay with the student's data everywhere. After the student is removed from the main table of student, then I do not want that student's data to be present in any other table. If I do find at any point of time that the student's data is present, even though the student is not with the institute, then that would be inconsistent, right? Similarly, in a banking example, I would want that the total of these two people's balance was X amount before the transfer and now should also be an X amount after the transfer because it just got deducted from one person and got credited to the other person.
There should be no point of time in which the sum is different from earlier. So, these kind of consistencies, rules can be present in your database and you are the one who is actually going to make sure that, you know, let's say some column should always be present, some column cannot be null, some things have only to be unique. For example, the student ID has to be unique, we cannot have any duplicate student ID. So, let's say there has been some transaction, a new student has been inserted, then I do not want that the consistency of unique student ID be violated.
So, all these things come under consistency. Now, what about isolation? Remember that any particular real-world database would run a lot of transactions concurrently. And, while doing these transactions, there could be intermediate state of each transaction that should be invisible from others. For example, somebody is typing out a comment in a social media website. Now, it may be that I am giving a feature to other people that, you know, somebody is typing out a comment, but then nobody should be able to see what comment the person is typing until and unless the person has already posted the comment.
Similarly, if a person is updating the database, I do not want other users to see the updates to the database until and unless the person has committed that transaction, the transaction is complete. This is the kind of isolation I want my database to have, that any other people, whatever they are doing with the database, should be invisible from the remaining users until and unless that transaction is complete. So, what would happen is that if we make sure that all transactions are fully isolated from each other, it appears that all these transactions are actually running one after the other instead of concurrently at the same time.
For example, the comments in any social media post come one after the other to any particular end user, right? Even though all these users might be typing at the same time, but whenever the comments are pushed on the post, they are always pushed one after the other and not together. This is the kind of isolation which we want in our application. This is the kind of isolation which we want in our database as well. And lastly, about durability, whenever a transaction is successful and it is committed, then it must persist on the disk in the sense that that data should be present in the database, in the hardware of the database.
So, let's say suddenly 2,000 people liked a particular photograph in a social media website. That information should be present on the disk so that the person who has posted, let's say, opens up their app in the evening, should be able to see that, oh, today 2,000 people liked my photograph. If the data was not durable, if it was not stored in the server's disk, then the data might just vanish away once the app refreshes in the user's mobile or in the user's computer. If you clear things from cache and you are actually not persisting it to the disk, to the actual hardware of the database, then you will lose that information.
And we do not want to lose the information that so many people actually liked that photograph because then the user experience is very bad, right? You have people recognizing your photograph, you have people recognizing your post, but that was never persisted on the disk and that was never shown to the original poster and that was never shown to the world in general. And that is a bad, bad user experience for people, right? We want durability to happen within our database. And this storage in the disk is not something which we want to undo.
That cannot be undone. So let's say there are 2,000 people liking it today. Tomorrow, nothing should happen that those 2,000 likes go away. So yes, every user has the capability to unlike a particular photograph or uncomment on a particular post. But all that data which was actually persisted on the disk should not suddenly vanish. Even in the event of any crash which may happen of the application or which may happen of the database system, we want that durability, we want that consistency. So I hope you are clear with all these 4 features, atomicity, isolation, consistency and durability.
Durability, simple that your data should persist on the disk. No crash should actually affect the database management system to keep your data secure. Consistency would make sure that all the rules which you have defined for your database and all the rules which you have defined for the data in the database is always there whether before a transaction or after a transaction. Isolation makes sure that all the transactions happening by multiple users are invisible from each other and only visible to other users once it is committed.
Atomicity makes sure that either the entire transaction is committed or nothing of it happens so that the user is not in a confused state about their activity in the application or in the database. We also want some other things from a database management system. For example, it would be great and it is in fact a very important requirement for modern systems that database management system does an automatic backup and recovery procedure for data. Just in case our database disk crashes or some corruption happens in our database, then we should be able to quickly backup our earlier data and we should be able to quickly recover any problems which would have happened to our data.
We also want to easily manage large amounts of data with a database management system. We want the DBMS to handle all the security aspects of our database for us. We do not want any illegal access to our data, any theft of our data, any corruption of our data from any hacker or any malware in a particular application. We want the database management system to handle all the security aspects for us. We also want to handle imports and exports of data easily using a DBMS. We want to handle interactions with web applications from DBMS.
For example, MySQL is an extremely good DBMS. But if it does not give me any capability of connecting from my web application code, let's say Java code, to MySQL, then I would not want to use it, right? But here's the thing. MySQL actually provides us very good capabilities to connect our web applications with the database. So that's why many people, many companies use MySQL. And that's the same with any other popular database management system because they give us such good interactions with web applications that we are able to easily use them for our web application and for all the other things which I'm mentioning right now.
A good DBMS would also provide a nice graphical user interface. For example, we have MySQL giving us different softwares like PHPMyAdmin or HeidiSQL or any other GUI-based system which would help us to deal with the database. The DBMS should also help provide easy maintenance of data. It should also help us reduce time by improving efficiency of anything we want to do with the data. It should also help us with data integrity and for controlling redundancy. For example, if we have our student data for our particular institute and we are storing the student's name.
Tomorrow, the student actually comes up with the request to change the student's name. Now, if we have the student's name stored in the database tables at more than one places, what would happen is we would actually have to make sure that our software actually updates the student's name in multiple tables. Or we would have to actually manually go and change the student's name in multiple tables. What if we forget to change it at one place? That would actually cause the data to not have integrity. Integrity means everywhere the data has to be same.
It has to be consistent. It has to not violate each other. There should be no inconsistency. There should be no difference in the data which we have of that particular student. So, wherever the student's name is present, there the student's name has to be so and so. If the student's name was John Smith and tomorrow the student says that I don't like the name John Smith and I've changed it to David Smith, it should be changed to David Smith at all the places. That student with the ID of let's say 12 is David Smith at one place and then John Smith at another place.
That should not happen. So, DBMS should also give us this capability that we are actually storing the student's name only at one place so that we avoid any kind of redundancy. What is the meaning of redundancy? It means that you are not repeating your data at multiple places. You only have it at one place. That is the meaning of avoiding redundancy. And if you avoid redundancy, you can achieve data integrity because you don't have to actually update it at multiple places. A DBMS should also facilitate data sharing.
Just the way we would want it to connect with our web application for data sharing or we would want to import or export data. We would want all those facilities from the DBMS. Also, with cloud data centers for modern web applications nowadays, for example, any web application might be having their data not in a particular hardware disk on a server. However, it would be present in a data center of a cloud provider like Amazon AWS or Google Cloud Platform or IBM Cloud Provider or any other cloud provider out there.
With such cloud data centers, the DBMS is capable of storing almost infinite data because they have a lot of space in their data centers and you can actually store a lot of data. And based on the amount of data which you want to store, they would charge you for a fee. And that is a characteristic of the database management system which is something which you would want. Probably, you want to store infinite amount of data, however much amount of data which your website would generate. You would want to store that and you want that DBMS to be capable of storing, retrieving and helping you with all of that data.
So, you would choose the particular DBMS in that way. So, for example, if I want to actually store my application on a cloud provider, I would choose a particular DBMS which would work very well with almost infinite data with very, very huge amount of data. And MySQL is something which actually helps us with a lot of these characteristics. In fact, all of them. That's why we have chosen to work with MySQL in this particular course. Remember, when I talk about data sharing using DBMS, it has to be for an authorized user only.
So, an authorized user can share the data among multiple users. But I do not want it for any unauthorized user. Also, when I talk about almost infinite data, I want to make sure that DBMS handles large amount of data on large disk space very efficiently. There could be some challenges as well. While working with a DBMS. For example, the size of a DBMS, you know, the entire software which we are going to use, that is huge. There is always a cost associated because we are using a good software.
So, we would obviously have to pay some cost to the DBMS provider. That is there. Then, obviously, there is some sort of complexity in actually using a database management system. Now, imagine you would keep all your institute's data in files, in registers, in physical hard copies of papers. That would be less complex for you to work with because you do not have to actually train your staff to use a DBMS system. They know how to use pen and paper and probably add entries to a register, etc.
But you would have to train people to use a DBMS system and that is the added complexity. But then, whatever benefits you get from a DBMS is something which you really want to make sure to use because we are heading towards modern systems. We are heading towards solving business challenges based on meaningful information from a lot of data. And that is what we are actually going to deal with in DBMS. And the best thing you can do to actually overcome any of the challenges of the DBMS is by choosing the right DBMS, being skilled in using that particular DBMS and being able to design it so that you always have efficient and accurate storage and retrieval of data.
You want to make sure when you are developing a large database, you want to install it properly, replicate the database properly, back it up always. And all this can be difficult and expensive. You definitely need a better hardware to handle load on a database server. Remember, whenever any web application goes down, it's almost always the first thing which is going down is the database server because the database server is the one which has the maximum load from the entire web application architecture. And you really need to make sure that you are handling the load on the database server very well.
So, you need a very good hardware for that. Also, if you have very high volume transactions, so you would definitely have longer processing times. For example, you want to book a flight ticket during New Year. And that's the time when maximum people are booking flight tickets. And you would realize that the response from the server is very slow during that time. Or any time, let's say, there is a sale and you are actually working with an e-commerce website. So, you would realize that, you know, if the website has not scaled its database properly, then you would have to actually have to wait for a longer processing time in the web application.
You don't want that. You should definitely try to scale your database properly so that there is no longer processing times during high volume transactions. And remember that security. Thank you. Security and integrity of a database is very, very critical. I mean, I cannot stress it enough. How important is security of a database? A business can completely get ruined just because their database was compromised or their data integrity was compromised. Let's say one day Facebook says that all the data of all the users is lost. So all your photographs of last so many years, which were stored in Facebook, you forgot to back it up on your PC.
All of that is lost. You would hate Facebook for that, right? Remember that security and integrity of the data is very, very critical for a database management system. And that is something which you have to make sure that your database management system is taken care of properly and you are taking care of that properly as well. Now, we actually talked so much about, you know, the challenges of a particular database. Now, why do I actually, you know, get so stressed about that? Why can't we actually put all of our information directly into a file system?
Just dump all the information in a particular file. Whenever you want to do it, you can actually have some efficient algorithms to search things. For example, you can dump all the information in a Word file and then, you know, use the Ctrl-F feature to search for certain items and then update it at, you know, those places. Why do we not do that? Why do we actually go ahead and use a DBMS instead in our web applications? One of the reasons is that we do not want redundant and inconsistent data, right?
We do not want data just scattered in multiple places. We do not have any organization of, we do not even remember where this data is. And every time we have to do a Ctrl-F, find it at each place and then update it. We do not want any of these problems. So, we use a database management system. Remember that if you have redundant data and if you have inconsistency problems in your data, this actually costs more storage. So, if you have, you know, one data stored at multiple places, you obviously would need higher storage, right?
And then, obviously, you would have higher access cost to that data because now, since the storage is so much, you have so much of data, your controller feature also, your find feature also, your search feature also is going to take a lot of time. And that adds to the cost of using that particular system for accessing data. Now, cost not necessarily means in the monetary sense. It also means in the time. If the user is taking more time to access data from your file system, that means your website is more costly for the user and the user would definitely switch over to an application which actually helps access data very fast in least amount of time.
Obviously, if you use a file system, you would have difficulty in accessing data because it is not well organized. Also, you would have no data isolation. Data may be scattered in various files and in different formats. You don't want all these problems. It may be that because of all this redundancy, you have integrity problems. Data is updated at some places but not at other places and then you have a whole big problem of integrity that you may have atomicity problems. That, you know, the user data got changed at some place but, you know, like for example, the user wants to actually pass on some information.
So, the information went from the user but never reached the destination. Let's say the user balance got deducted but never incremented on the receiver's end. You do not want all these problems in a particular file system. That's why you would go for a database management system. Also, there are multiple people using the same data at the same time, right? It may be that your file system is not able to handle that properly. So, you would have concurrent access anomalies. It may be that multiple people are updating a particular file and one user wants to add a particular comment at one place and you actually have two comments coming in and, you know, both comments get mixed up and then you have a whole gibberish thing over there.
Also, you may have security problems while working with file system which a DBMS system would take care of. So, because of all these problems, we do not go for a file system, we go for a database management system which would help us overcome all these problems.