Load balancing in system design interviews: algorithms, health checks and failover

A load balancer is the component that decides which server receives each request once a system runs more than one instance. Interviewers rarely accept simply adding one; they probe which routing algorithm is used, what the balancer inspects, and how it detects a failed backend. The article explains that the balancer provides scalability and availability but introduces its own concerns, including sticky sessions and the risk of becoming a single point of failure. It frames load balancing as a recurring probe across many design problems, where the hard behaviour appears under failure rather than in the diagram. Strong answers are judged on choosing an algorithm suited to traffic patterns and on handling health checks and the balancer's own redundancy.