Home/Writings/Engineering

Observability vs Logging

Why logs are useful, but not enough to understand a running system.

Figure 1A working loop for technical inquiry: observe the system, reason about its constraints, and build to test the model.

Logs are valuable because they preserve narrative. A well-placed log line can explain a decision, expose an input, or show the exact branch a request took through a service.

Observability is wider than that. It asks whether you can answer new questions about the system without shipping new code. Metrics, traces, structured logs, and domain-specific events each cover a different part of that problem.

The trap is collecting everything and understanding nothing. Useful observability starts from the questions you expect to ask during an incident: where is time going, what changed, who is affected, and which dependency is shaping the failure?

Continue reading
Systems Design · 12 min read

The Hidden Cost of Abstractions

Why every layer leaks eventually, and why backend engineers should care.

Read note
Distributed Systems · 14 min read

Raft Consensus and Failure

A look at why the easy part of consensus is the normal case.

Read note
Distributed Systems · 8 min read

Emergence in Complex Systems

Why distributed behavior can surprise you even when every local rule is simple.

Read note