The Cache Answers First

Core Question

Core Question

Why do most DNS queries never leave the resolver, and what exactly decides hit versus miss?

Outcome

Outcome

By the end of this session, the learner should be able to:

• explain why the cache is consulted before any network step
• name the key a cache entry is stored under
• explain why the same name with a different type is a miss
• describe what happens to an entry the moment it expires

Read Order

Read Order

1. Read CacheEntry
2. Read ResolverCache
3. Read entry_is_expired()
4. Read lookup()
5. Read store()
6. Run examples/dns/session_04_walkthrough.py

Read It Like Code

Read It Like Code

CacheEntry(
    records,
    stored_at,
    ttl,
)

Fields That Matter

Fields That Matter

Decision Flow

Decision Flow

key not in cache            -> Miss   (go to the network)
entry past stored_at + ttl  -> Expired (delete, then go to the network)
otherwise                   -> Hit    (answer locally)

Reading Lens

Reading Lens

The important move in this session is to stop thinking of the cache as an optimization detail and start asking:

• what key was this answer stored under?
• what time did lookup() run, relative to stored_at + ttl?
• did this query touch the network at all?

Toy Model Boundary

Toy Model Boundary

Real caches store negative answers, partial chains, and per-record TTLs, and they cap TTLs with local policy. This lesson keeps one entry per question key so the hit/miss/expired branching stays readable.

Time is an integer passed in as now. Making the clock an argument instead of calling a clock inside is what makes every scenario in the walkthrough reproducible.

Code Landmarks

Code Landmarks

question_key() (from Session 01)

The cache key is the question identity from Session 01, unchanged. Different type, different key, different entry.

entry_is_expired()

One comparison: now >= stored_at + ttl. Expiry is arithmetic, not a background process.

lookup()

The main reading target. Three outcomes, and the Expired branch also deletes — an expired entry is treated as if it were never there.

store()

A network answer becomes a cache entry stamped with arrival time. Nothing else about the resolver changes.

Failure Questions

Failure Questions

Use the source file to answer these:

1. Why is lookup() for WWW.Example.COM. a hit after storing www.example.com?
2. Why is lookup() for the same name with qtype="AAAA" a miss?
3. At exactly now == stored_at + ttl, is the entry alive or expired? Which comparison decides?
4. After an Expired lookup, what does the very next lookup for the same key return, and why?
5. Why does lookup() take now as a parameter instead of reading a clock?

Walkthrough

Walkthrough

Run this:

PYTHONPATH=src python3 examples/dns/session_04_walkthrough.py

The walkthrough stores one answer and shows hit, miss, and expiry purely by moving now.

Done When

Done When

The learner can say all of the following without looking at notes:

• "The cache is keyed by the full question, so name alone is not enough."
• "Hit, miss, and expired are the only three outcomes, and expired behaves like miss plus a delete."
• "Most queries are answered by arithmetic on stored_at + ttl, not by the network."

References

References

• RFC 1034 Section 5.3.1
• RFC 1035 Section 7.4
• RFC 2308 Section 5 (caching, including negative caching this toy skips)

Continue

Keep reading DNS