Understanding Durability in ACID

Arpit Bhayani

entrepreneur, educator, and tinkerer

After discussing the “A”, the “C”, and the “I”, it is time to take a look at the “D” of ACID - Durability.

Durability seems to be a taken-for-granted requirement, but to be honest, it is the most important one. Let’s deep dive and find why it is so important? How do databases achieve durability in the midst of thousands of concurrent transactions? And how to achieve durability in a distributed setting?

What is Durability?

In the context of Database, Durability ensures that once the transactions commit, the changes survive any outages, crashes, and failures, which means any writes that have gone through as part of the successful transaction should never abruptly vanish.

This is exactly why Durability is one of the essential qualities of any database, as it ensures zero data loss of any transactional data under any circumstance.

A typical example of this is your purchase order placed on Amazon, which should continue to exist and remain unaffected even after their database faced an outage. So, to ensure something outlives a crash, it has to be stored in non-volatile storage like a Disk; and this forms the core idea of durability.

How do databases achieve durability?

The most fundamental way to achieve durability is by using a fast transactional log. The changes to be made on the actual data are first flushed on a separate transactional log, and then the actual update is made.

This flushed transactional log enables us to reprocess and replay the transaction during database reboot and reconstruct the system’s state to the one that it was in right before the failure occurred - typically the last consistent state of the database. The write to a transaction log is made fast by keeping the file append-only and thus minimizing the disk seeks.

Durability in ACID

Durability in a distributed setting

If the database is distributed, it supports Distributed Transactions, ensuring durability becomes even more important and trickier to handle. In such a setting, the participating database servers coordinate before the commit using a Two-Phase Commit Protocol.

The distributed computation is converged into a step-by-step process where the coordinator communicates the commit to all the participants, waits for all acknowledgments, and then further communicates the commit or rollback. This entire process is split into two phases - Prepare and Commit.


Courses I teach

Alongside my daily work, I also teach some highly practical courses, with a no-fluff no-nonsense approach, that are designed to spark engineering curiosity and help you ace your career.

System Design Masterclass

A no-fluff masterclass that helps experienced engineers form the right intuition to design and implement highly scalable, fault-tolerant, extensible, and available systems.

Details →

System Design for Beginners

An in-depth and self-paced course for absolute beginners to become great at designing and implementing scalable, available, and extensible systems.

Details →

Redis Internals

A self-paced and hands-on course covering Redis internals - data structures, algorithms, and some core features by re-implementing them in Go.

Details →

Arpit Bhayani

Arpit's Newsletter

CS newsletter for the curious engineers

❤️ by 90000+ readers

If you like what you read subscribe you can always subscribe to my newsletter and get the post delivered straight to your inbox. I write essays on various engineering topics and share it through my weekly newsletter.

Writings and Learnings

Knowledge Base



Arpit's Newsletter read by 90000+ engineers

Weekly essays on real-world system design, distributed systems, or a deep dive into some super-clever algorithm.