Proactive scalability is used to remove human intervention. It is possible to create a latency metric and measure the requests per second in a time interval, or perhaps the CPU use percentage, in order to create an alarm and automatically execute a script that moves the virtual IP to a more efficient and robust node. Using this same rationale, we can make a fully-autonomous system that adds redundant copies of our web server, and a load balancer that homogeneously distributes requests, increasing the overall system capacity that adapts to different traffic patterns. This is the core responsibility of Auto Scaling, a service that will be discussed in depth in the following chapters.