How would you design a Julia runtime observability system for production?

You would combine profiling hooks, allocation tracking, distributed tracing, and compile-time introspection.

A production observability system for Julia must track allocations, method dispatch events, compilation latency, and runtime performance metrics. Tools like Tracy.jl or custom instrumentation hooks can capture function-level latency. Distributed systems require tracing across processes and workers to understand bottlenecks.