Instagram Scales on Python for 2 Billion Daily Users (2024)

Instagram, one of the most famous Social Media apps of our time, sharing enormous amount of photos, videos and media content, STILL RUNS ON PYTHON..! Everyone of you might have heard numerous different things about Python -

• Python is too slow.
• Python does not scale for millions of API requests
• It is not concurrent / multi threaded
• No company uses Python for scale. You can only use JAVA or Golang.

Well, guess what..!! Instagram, one of our favourite social media platform still works on a Monolithic Python's Django codebase..! Obviously, there are so many more services which could be using so many more frameworks and languages, but over here I'll talk about the main Backend Service powering Instagram's App - feed, content, media service.

Instagram's Scale

Now, with that scale, back in 2016, they also had about 4.2 Billion Likes being sent every single day..! Does that blow your mind?

Now, to handle this scale for every single day, Instagram has a scale of 10's of thousands (or even more 100's) of Django Servers running, scattered across the world, fault tolerant and so much more..!

Instagram's History with Django

Instagram started with Django back in 2010 as their go to tech stack. Why? Because it's super easy to setup and took only 2 weeks to start and a couple of months to reach a million users..! A single Django server..!

Only in 2011 Summer, Instagram had about 5M users with all of its data being stored in PostgreSQL and using Django's ORM. As and when Instagram grew, they saw problems with Django's ORM, the complexity of vertical partitioning (available in Django at the time) and so many more things.

They started out with Horizontal Partitioning support in Django, cleaning but multiple ORM transaction support things and so much more.!

The Famous Django Celery Tech Stack

Now, anyone who works in Python's backend may have already used Django and Celery together..! They are an absolute love, and async processing without Celery is just unheard about..!

Celery is one of the best task Scheduler and Executor available in Python which comes out of the box with many many essential features. Plus it with RabbitMQ or Redis, you are good to go with any scale of Async Processing..!

Now the reason, why you see this stack in every single Python based product is because this is the original Instagram's Stack which they had used (still use) and you obviously know about Instagram's scale, don't you?

Is Instagram the only company using Python?

Why Python?

Simple -

• Fast developer productivity
• Scalable
• Ease to ship and maintain
• Enormous amount of features and libraries

Reading Material

I hope you got the basic essence of this, from a very simplistic view. If you like this article, do Like ❤️, Share 🔁 and Comment 💬 and don't forget to subscribe ✅ to this newsletter..!