During a 60-minute HackerRank online interview, I was given a single large system-design problem divided into five parts: to build a server load-balancing system.

My task was to design data structures and algorithms that could maintain a pool of servers and distribute client connections according to a chosen balancing policy such as least-loaded, round-robin, or consistent hashing.

A key requirement was to implement a shutdown operation, which shuts down a specified server by its ID and then reassigns all existing connections on that server to other servers using the same logic as the normal CONNECT operation, ensuring consistent and balanced redistribution.

Over the course of the exercise, I needed to think through the full lifecycle—tracking servers and connections, handling edge cases like the absence of available servers, and optimizing for both time and space efficiency—while writing working code for each of the five progressive parts.

Design a server load-balancing system.

You need to build the core data structures and logic that manage a pool of servers and distribute client connections across them according to some rules (not all rules are written out here, but think along the lines of “CONNECT” requests and balancing algorithms).

One of the specified operations is called tdown:

tdown can shut down a given server identified by an ID or sequence number.

When a server is shut down, any existing client connections on that server must be re-assigned to other servers.

The re-assignment must follow the same rules as a normal CONNECT operation—i.e., whatever balancing strategy you defined for new connections (e.g., least-loaded, round-robin, consistent hashing) must be used to redistribute those connections.