The Load Balancing problem assigns N jobs with varying processing loads to M identical processors, minimizing the maximum load on any processor (known as the makespan).

Sort jobs in descending order of size. Assign each job to the processor that currently has the minimum accumulated load (Graham's Longest Processing Time heuristic).

While load balancing is NP-hard, the greedy LPT choice guarantees a 4/3-approximation ratio of the optimal schedule makespan.

Proof by approximation bounds: sorting descending prevents large jobs from being scheduled late, limiting the maximum deviation from the optimal average load to at most 33%.