open atlas
↑ Back to track
System Design Foundations SD · 03 · 06

Traffic & edge: free-recall review

Free-recall prompts across the traffic unit. Answer each in your own words first — including the tradeoffs and the arithmetic — then reveal the model answer and compare.

SD Senior ◷ 14 min
Level
FoundationsJuniorMiddleSenior

Retrieval beats re-reading. For each prompt, say or write a full answer from memory — including the tradeoffs and the arithmetic — before you open the model answer. The effort of reconstructing the L4/L7 split, the algorithms, the gateway’s blast radius, and the hit-ratio economics is what makes them stick.

Reconstruct the unit’s spine without looking back: where a balancer reads the request and which algorithm protects the tail, what a gateway adds and concentrates, when a mesh fits, and how a CDN’s hit ratio, TTL, and invalidation work.

Recall before you leave
  1. 01
    Contrast L4 and L7 load balancing, and name the algorithms from blind to load-aware.
  2. 02
    What is consistent hashing for, and why not plain hash-modulo-N?
  3. 03
    Distinguish forward proxy, reverse proxy, and API gateway, and state the gateway's core tradeoff.
  4. 04
    When do you reach for a service mesh, and how does north-south vs east-west split?
  5. 05
    Walk a CDN's two benefits, the hit-ratio economics, push vs pull, and what not to cache.
Recap

If you could reconstruct each answer from memory, you hold the unit’s spine: a balancer reads requests at L4 (fast, protocol-blind) or L7 (content-aware, at a CPU cost), and its algorithm decides the tail — load-aware least-connections/EWMA route around a slow node, while consistent hashing preserves cache/shard locality. A gateway is a reverse proxy that centralizes auth/rate-limit/routing/TLS, a consistency win that becomes a SPOF/bottleneck unless run redundant and thin; a service mesh handles east-west concerns in per-instance sidecars. A CDN trades on hit ratio (non-linear: 90% → 95% halves origin load), gets content via pull (default) or push, and its hard problem is invalidation (purge active, TTL passive, content-hashed URLs avoiding it), with origin shielding against the herd and a hard rule never to cache per-user or must-be-correct data on a shared edge. The thread: at this depth, traffic and edge are reasoning you do up front, not behavior you discover in production.

Something unclear?

Ask a question about this lesson. Questions are anonymous and go straight to the author to make the lesson better.

shortcuts expand
search
K
prev piece
k
next piece
j
cycle tier
t
this menu
?
sources3
expand
  1. 01
  2. 02
  3. 03

Trademarks belong to their respective owners. Editorial reference only.