Hacker News Reader

This only works in very predictable workloads. Once your average input rate becomes slightly larger than consumption rate your system is screwed. Over provision ensures spikes can be processed along with average rate in a general increased load state for some time.

View on HN

True, my example is an oversimplification. In reality you want to account for rapidly and slowly changing demands. This can be solved with elasticity of systems or by overprovisioning. You can get deep, with time series analysis, low pass filtering and various metrics, such as quantile distributions… however, saying zero queue length is desirable is a gross oversimplification.

Or use load shedding, in which case you want your queue nonempty to take advantage of negative spikes.

Reply on news.ycombinator.com