Engineering Is About Intentional Sacrifice: Lessons from Jonas Bonér
Every architect has that moment where they realize that "best practice" is just a fancy term for "least-worst trade-off."
For the longest time, I hunted for the silver bullet — the perfect database, the perfect cache, the perfect distributed pattern. After spending serious time digging into Jonas Bonér's work on system stability, I have realized (again) that engineering is not about perfection. It is about intentional sacrifice.
I have been deep-diving into his notes on scalability and patterns, and it has been a total perspective shift. Here are the moments from that study session that are changing how I review my implementations.
Performance vs. Scalability: The One User Rule
Performance: If your app is slow for just you (one user), that is a code or query optimization problem. Fix your logic.
Scalability: If your app flies for one user but crashes during a traffic spike, that is an architecture problem. Fix your infrastructure.
They sound similar but demand completely different solutions.
The Latency vs. Throughput Tug-of-War
Think of it like a highway. Latency is how long it takes your car to reach the destination. Throughput is how many cars the highway can handle in an hour.
The goal is not the impossible. It is maximal throughput while keeping latency acceptable for the user.
The Reality of the CAP Theorem
In distributed systems, the network will break — that is the P in CAP. When it does, you have to choose your poison:
- CP (Consistency + Partition Tolerance): The system locks up rather than show you potentially stale data. Accuracy over uptime.
- AP (Availability + Partition Tolerance): The system stays live, but you might see slightly outdated information. Uptime over immediate accuracy.
Why This Matters
If you are working on a traditional RDBMS, you are in a cozy bubble of ACID properties — Atomic, Consistent, Isolated, Durable — where everything is tidy. But once you move into distributed systems, you stop solving problems and start managing trade-offs. Knowing which one to sacrifice is what separates a senior engineer from the rest.
If you want to nerd out on these patterns, check out Jonas Bonér's slides on scalability, availability, and stability patterns — they are gold:
🔗 https://www.slideshare.net/slideshow/scalability-availability-stability-patterns/4062682
Huge credit to Jonas Bonér for breaking these complex concepts down into actual, usable patterns. His work is mandatory reading for anyone scaling a platform. More snippets from his slides to come.