Diagnose Slow Network Throughput Including Containers

A service has unexpectedly low network throughput, including when running inside a container. Explain how you would isolate the bottleneck.

Constraints & Assumptions

The question expects a low-throughput investigation and reasoning about containers. No specific protocol, topology, or provider is assumed. Separate application processing from transport and infrastructure constraints.

Clarifying Questions

Is the issue throughput, latency, connection establishment, or sporadic failure? Which endpoints and routes are impacted? Does the identical workload perform differently on the host versus inside the container?

What a Strong Answer Covers

A strong answer covers layered comparisons, packet loss and retransmission behavior, bandwidth-delay effects, host vs. container limits, MTU and path concerns, and experiment-driven evidence.

Follow-up Questions

How might a CPU bottleneck mimic a network limit? Why can a single connection fail to fully utilize available bandwidth? What would you check before attributing blame to the container network overlay?

Overview: Investigate low network throughput across application, transport, host, and container layers by using controlled comparisons, retransmissions, limits, and path behavior.