Distribute traffic across instances with the right algorithm, health checks, and draining so capacity is used and failures are absorbed. Use when scaling a service horizontally or diagnosing uneven load.
