Module 03 · Load balancing · Drill

Load balancing drill

Scenario prompts that check whether load balancing is usable under pressure. Answer out loud, then compare.

4 prompts20 minNew

0 of 4 graded

Prompt 1
A stateless API runs on eight identical servers behind a round-robin balancer. Engineers notice p99 latency is poor even at modest load, and one server is always hotter than the others. Requests range from 5 ms lookups to 10-second exports. Diagnose and fix.
Say your answer out loud first. Thirty seconds is enough.
Prompt 2
During every deploy, about 0.5 percent of requests fail with 502 for roughly thirty seconds. The pool has health checks every ten seconds with a three-failure threshold. Explain why and lay out a zero-downtime deploy sequence.
Say your answer out loud first. Thirty seconds is enough.
Prompt 3
A product serves users in North America and Europe from one US region. Leadership wants EU users to get EU-local latency and wants the service to survive the loss of a region. Design the traffic layer and state the failover behaviour precisely.
Say your answer out loud first. Thirty seconds is enough.
Prompt 4
A colleague proposes health checks that call the database, the cache, the search cluster and the email provider, so the balancer 'only sends traffic to fully working instances'. Argue for or against, and specify what the readiness check should contain.
Say your answer out loud first. Thirty seconds is enough.