What are ACID transactions?


"Programming should be about transforming data"

— Programming Elixir 1.3 by Dave Thomas

As developers, we are interacting oftenly with data, whenever handling it from an API or a messaging consumer. To store it, we started to create softwares called relational database management system or RDBMS. Thanks to them, we, as developers, can develop applications pretty easily, without the need to implement our own storage solution. Interacting with mySQL or PostgreSQL have now become a commodity. Handling a database is not that easy though, because anything can happen, from failures to concurrency isssues:

  • How can we interact with datastores that can fail?
  • What is happening if two users are updating a value at the same time?

As a database user, we are using transactions to answer these questions. As a developer, a transaction is a single unit of logic or work, sometimes made up of multiple operations. It is mainly an abstraction that we are using to hide underlying problems, such as concurrency or hardware faults.

ACID appears in a paper published in 1983 called “Principles of transaction-oriented database recovery” written by Theo Haerder and Andreas Reuter. This paper introduce a terminology of properties for a transaction:

Atomic, Consistency, Isolation, Durability


Atomic, as you may have guessed, atomic represents something that cannot be splitted. In the database transaction world, it means for example that if a transaction with several writes is started and failed at some point, none of the write will be committed. As stated by many, the word atomic could be reword as abortability.


You will hear about consistency a lot of this serie. Unfortunately, this word can be used in a lot of context. In the ACID definition, it refers to the fact that a transaction will bring the database from one valid state to another.


Think back to your database. Were you the only user on it? I don’t think so. Maybe they were concurrent transactions at the same time, beside yours. Isolation while keeping good performance is the most difficult item on the list. There’s a lot of litterature and papers about it, and we will only scratch the surface. There is different transaction isolation levels, depending on the number of guarantees provided.

Isolation by the theory

The SQL standard defines four isolation levels: Serializable, Repeatable Read, Read Commited and Read Uncommited. The strongest isolation is Serializable where transaction are not runned in parallel. As you may have guessed, it is also the slowest. Weaker isolation level are trading speed against anomalies that can be sum-up like this:

Isolation level dirty reads Non-repeatable reads Phantom reads Performance
Serializable ๐Ÿ˜Ž ๐Ÿ˜Ž ๐Ÿ˜Ž ๐Ÿ‘
Repeatable Read ๐Ÿ˜Ž ๐Ÿ˜Ž ๐Ÿ˜ฑ ๐Ÿ‘๐Ÿ‘
Read Commited ๐Ÿ˜Ž ๐Ÿ˜ฑ ๐Ÿ˜ฑ ๐Ÿ‘๐Ÿ‘๐Ÿ‘
Read uncommited ๐Ÿ˜ฑ ๐Ÿ˜ฑ ๐Ÿ˜ฑ ๐Ÿ‘๐Ÿ‘๐Ÿ‘๐Ÿ‘

I encourage you to click on all the links within the table to see everything that could go wrong in a weak database!

Isolation in Real Databases

Now that we saw some theory, let’s have a look on a particular well-known database: PostgreSQL. What kind of isolation PostgreSQL is offering?

PostgreSQL provides a rich set of tools for developers to manage concurrent access to data. Internally, data consistency is maintained by using a multiversion model (Multiversion Concurrency Control, MVCC).

Concurrency Control introduction

Wait what? What is MVCC? Well, turns out that after the SQL standards came another type of Isolation: Snapshot Isolation. Instead of locking that row for reading when somebody starts working on it, it ensures that any transaction will see a version of the data that is corresponding to the start of the query. As it is providing a good balance between performance and consistency, it became a standard used by the industry.


Durability ensure that your database is a safe place where data can be stored without fear of losing it. If a transaction has commited successfully, any written data will not be forgotten.

That’s it?

All these properties may seems obvious to you but each of the item is involving a lot of engineering and researchs. And this is only valid for a single machine, the distributed transaction field is even more complicated, but we will get to it in another blogpost!

Thank you for reading my post! Feel free to react to this article, I am also available on Twitter if needed.