TTL-Based Caching

Every DNS record carries a TTL in seconds. A resolver MUST NOT serve cached records past their TTL.

Cache structure (resolver):

Real-world resolvers also do:

• Negative caching (RFC 2308): cache NXDOMAIN responses for the SOA's minimum field.
• TTL flooring/ceiling: cap TTL at a maximum (e.g., 1 hour) to allow faster refresh, or floor at a minimum.
• Stale-while-revalidate: serve stale data immediately, refresh in background. Improves apparent latency.
• Pre-fetching: when a popular record's TTL is about to expire, refresh proactively.

Cache invalidation: there isn't any in DNS. If you change a record, you must wait for TTLs to expire across all caches. That's why DNS migrations involve "lower TTLs ahead of the change, swap, restore TTLs".

For this lesson: implement TTL-based caching with simulated time.