Time to live

So where does the data go?

It just holds the fetched metadata in an ephemeral HTTP cache tier, like in memory storage or a Redis cluster, and that cache is naturally bounded by TTL headers.

OK. Criterion two, security posture and attack surface.

This is the big one.

If a developer wants to decommission an app, they just delete the JSON file from their web server.

And that's it? That's it.

Once the authorization server's cache, TTL, expires, usually within 24 hours, the client ceases to exist globally.

Zero orphan records.

Never make a DNS change in a hurry you didn't plan.

And before a plan changed, like a host move, lower the TTL a day ahead so the switch lands faster.

Habit is the documenter's habit, and it matters more in DNS than almost anywhere else.

because a mistyped record takes down real services, and there is no undo button.

So when an LLDP advertisement comes in, they're going to grab onto it and store it.

Well, how long are they going to store it? They're going to store it for a TTL value, time to live value.

TTLs are something we run into all the time in networking where here's some information I'm going to give you.

I want you to keep it.

But I want you to keep it forever.

And after an hour, I want you to make that go away.

Just delete that information because it's probably no good anymore.

And so you see TTLs, again, in all kinds of different things in networking.

Fix it by using fully qualified names with a trailing dot for external calls, or override ndots in dnsconfig for the workloads that need it.

Tune the core DNS cache plugin's success and denial TTLs deliberately.

Don't just accept the defaults.

Because negative caching is what protects you from repeatedly re-querying names that don't exist.

Another point of confusion is the cache's lifespan.

Prompt caches have a time to live, or TTL, which is often short by default, such as five minutes.

You can't assume a prompt cached in the morning will still be available in the afternoon if it hasn't been used recently to reset that timer.

The exam might present a scenario where caching seems to fail intermittently, and the cause could be an expiring TTL or subtle changes to the prompt prefix.

To remember the core principle, use this phrase: cache the constant, cut the cost.

This mental shortcut reminds you that the unchanging constant parts of your prompt are what you should target for caching to achieve the primary benefits of lower latency and reduced operational expenses.