What Is Autoscaling?

Load balancing has a major blind spot—it can handle massive amounts of traffic and users, but it can't deal well with fluctuations and spikes.

Building a scalable web application with only load balancing means that we are either spending way too much on an oversized infrastructure to deal with potential spikes, or we risk degraded service on a sudden influx. Both scenarios are not desirable.

Autoscaling to the rescue

We can define autoscaling as the process of automatically adjusting the amount of deployed resources based on load metrics. Autoscaling allows us to dynamically adjust our infrastructure based on current or predicted load. Therefore, we don't overspend on resources in normal load scenarios, but we are able to adjust in case of sudden spikes by quickly adding resources.

Load metrics

Load metrics are measurable units that represent the load on our resources and in the best case, accurately represent the bottleneck of our infrastructure.