Topic
Definition
Redis is an in-memory key-value data store commonly used as a cache, session store, or rate-limit counter that trades some durability for very low read and write latency.
In simpler words
Redis is a fast, temporary place to remember answers so Nest does not have to ask Postgres again right away.
No install is required this week. The goal is judgment: when caching solves a real, measured problem, and when it just adds a second source of truth to babysit.
Definition
Caching is justified when a read is expensive or frequent relative to how often its underlying data changes, and slightly stale results are acceptable for a bounded window.
In simpler words
Cache values that are read a lot, change rarely, and can be a few seconds old without causing harm.
Good candidates: a ticket-status dashboard summary hit on every page load, a rarely-changing settings blob, a rate-limit counter, or a session lookup.
Bad candidates: a single ticket a user is actively editing, anything where staleness causes a wrong authorization decision, or data that has never been measured as slow.
Ask what happens if this value is ten seconds stale. If the answer is nothing bad, it is a caching candidate. If the answer touches money, auth, or inventory, be careful.
async getTicketCounts() {
const cached = await this.redis.get('tickets:counts');
if (cached) return JSON.parse(cached);
const counts = await this.ticketsRepo.count();
await this.redis.set('tickets:counts', JSON.stringify(counts), 'EX', 30);
return counts;
}Cache-aside checks Redis first, falls back to Postgres, then repopulates the cache with a short TTL.
Definition
Cache invalidation is the deliberate act of removing or refreshing a cached value when its underlying data changes, and a cache failure mode describes how the system should behave if the cache is unavailable or wrong.
In simpler words
Decide up front how a cached value gets cleared, and what happens if Redis is down or returning stale data.
Two invalidation styles exist: TTL expiry, which is simple and always slightly stale, and explicit invalidation on write, which is fresher but adds more code paths to keep correct.
A cache should be allowed to fail. If Redis is unreachable, the endpoint should fall back to Postgres rather than return a server error. Treat a cache as an optimization, never a dependency the API cannot live without.
A common bug is caching per-user data under a shared key, which leaks one user data to another. Always include the tenant or user id in the cache key.
Test
At least 10 questions — mix of concept, syntax, practical, and logic. Score ≥ 80% (enforced by the API) to save progress.
Checking your session…
10 questions · concept 3 · syntax 2 · practical 3 · logic 2