DNS, measured properly.

The idea

A few lookups are not enough to choose a DNS resolver. Query order, caching, failures and the network path all affect the result.

What it does

• Queries resolver IP addresses directly, using the same domains across resolvers before moving to the next domain.
• The Bash implementation reads query timing from dig. The Rust implementation uses Tokio and hickory-proto for UDP queries and wall-clock timing.
• Reports success rates, timeouts, median and tail latency, alongside per-query CSV and structured summaries.
• Includes configurable duration, concurrency, resolver lists and domain lists.