Canva's S3 Session Revocation Architecture Explained – Scaling 100M+ Sessions! (2026)

The tech world is obsessed with scale—how to grow without breaking, how to innovate without collapsing under its own weight. But few stories capture this tension better than Canva’s recent overhaul of its session revocation system. Let me tell you, this isn’t just about cookies and databases. It’s about the invisible war between security, performance, and the sheer chaos of managing hundreds of millions of active sessions. And what Canva did? It redefined the rules of the game.

Let’s start with the elephant in the room: session revocation. It’s the digital equivalent of cutting off a thief’s access to your house while they’re still inside. But when you’re handling hundreds of millions of sessions, the traditional methods—like querying a centralized database—become a nightmare. Canva’s engineers faced a choice: stick with the status quo and risk system-wide slowdowns, or reimagine the entire architecture. They chose the latter.

Here’s what I find fascinating: they didn’t just swap out one database for another. They used Amazon S3—a tool most people think of as a file storage solution—as the backbone of their revocation system. How? By treating S3 as a distributed, immutable ledger of revocation events. Each revoked session becomes a 16-byte binary record, stored in 30-minute chunks. Gateways then download these chunks on demand, process them in memory, and discard anything older than 12 hours. The result? An 87.5% reduction in memory usage for their revocation cache. That’s not just efficiency—it’s a revolution in how we think about state management.

But let’s not get ahead of ourselves. The real genius here lies in the trade-offs. Canva’s engineers knew that frequent token refreshes would bloat databases and create bottlenecks. So they opted for a system that avoids constant database queries, instead relying on asynchronous workers to merge revocation data into S3 chunks. This means gateways don’t have to wait for a database response—they just check the latest S3 chunk and move on. It’s a subtle but profound shift: moving from reactive validation to proactive preparation.

Now, let’s talk about the Reddit debate. One commenter suggested, 'Just use a refresh token scheme.' Sounds simple, right? But here’s the catch: short-lived access tokens require constant database checks during refreshes. If your database is down, your entire system grinds to a halt. Canva’s approach sidesteps this by decentralizing the revocation state. The database now acts as a secondary layer, not the primary one. This isn’t just about speed—it’s about resilience.

What makes this particularly fascinating is the cultural shift it represents. Engineers are no longer designing systems around databases as the single source of truth. Instead, they’re treating distributed storage as a first-class citizen. S3 isn’t just for storing images anymore; it’s a coordination primitive, a way to distribute state across a network without relying on a single point of failure. This feels like the early days of microservices, where people realized that monolithic databases were the real bottleneck.

And let’s not forget the human element. Canva’s engineers had to convince themselves—and others—that this approach would work. Testing it on real infrastructure was critical. They couldn’t just rely on theory; they had to see if 2,000 revocations per second could be processed without a hitch. The fact that they succeeded suggests a broader trend: companies are becoming more comfortable with experimentation at scale. This isn’t just about solving a technical problem—it’s about building confidence in the unknown.

Looking ahead, this architecture raises a deeper question: How many other systems are built on outdated assumptions about state management? Canva’s approach could inspire a wave of rethinking in how we handle authentication, logging, and even real-time data synchronization. Imagine a world where every service treats storage as a distributed, immutable resource. It’s not just about scalability—it’s about reimagining the very fabric of digital infrastructure.

So the next time you log into Canva, remember: behind that seamless experience is a battle of ideas, a clash between old paradigms and new possibilities. And if you take a step back, you’ll realize this isn’t just about one company’s engineering win. It’s a glimpse into the future of how we build systems that can grow without breaking.

Canva's S3 Session Revocation Architecture Explained – Scaling 100M+ Sessions! (2026)
Top Articles
Latest Posts
Recommended Articles
Article information

Author: Amb. Frankie Simonis

Last Updated:

Views: 5850

Rating: 4.6 / 5 (76 voted)

Reviews: 91% of readers found this page helpful

Author information

Name: Amb. Frankie Simonis

Birthday: 1998-02-19

Address: 64841 Delmar Isle, North Wiley, OR 74073

Phone: +17844167847676

Job: Forward IT Agent

Hobby: LARPing, Kitesurfing, Sewing, Digital arts, Sand art, Gardening, Dance

Introduction: My name is Amb. Frankie Simonis, I am a hilarious, enchanting, energetic, cooperative, innocent, cute, joyous person who loves writing and wants to share my knowledge and understanding with you.