In the past when I thought about big-scale projects, my mind would usually wonder towards a system-design type of a mental exercise.

• What is the nature of the load, is it constantly high or changes dynamically based on time or season? Do we need infrastructure elasticity?
• Are our backend services prepared for scaling horizontally? How do we propagate information about sessions?
• How do we handle the databases pressure, do we need replicas or maybe even sharding and conflict resolution mechanisms?

And so on. I would rarely think about the frontend-consequence of a large scale system. I was lucky enough to experience this problem in the last couple of months!

The Scale

I’m talking about quite an impressive system, backend servers billions of unique user activities every month. Out of that, public events usually transfer into dozens or hundreds of updates every minute on the frontend.

Performance

It won’t be a surprise to say that such systems demand a high level of SPA performance awareness. What does it mean in practice:

• Working with websockets & optimizing their contract
• Optimizing usage of state management tools with websockets
• Minimizing React re-renders cascades
• Monitoring for memory leaks which accumulate over time