Intelligent autoscale of services

A method for performing an autoscaling operation for a plurality of servers executing on different physical machines that, together, provide a distributed service, the method comprising:

• accessing operational data associated with the plurality of servers that provide the distributed service;
• formulating, for a particular later time, a predicted load on the plurality of servers to provide the distributed service based on the operational data;
• estimating a capacity metric for providing the distributed service with the plurality of servers, while meeting a set of performance criteria, wherein the capacity metric is estimated based on historical operational data associated with the plurality of servers and a number of servers in the plurality of servers;
• and
• performing an autoscale operation to adjust a number of servers in the plurality of servers to provide better performance for the distributed service in view of the predicted load.

A non-transitory machine readable medium storing a program for execution by at least one processing unit for performing an autoscaling operation for a plurality of servers executing on different physical machines that, together, provide a distributed service, the program comprising sets of instructions for:

• accessing operational data associated with the plurality of servers that provide the distributed service;
• formulating, for a particular later time, a predicted load on the plurality of servers to provide the distributed service based on the operational data;
• estimating a capacity metric for providing the distributed service with the plurality of servers, while meeting a set of performance criteria, wherein the capacity metric is estimated based on historical operational data associated with the plurality of servers and a number of servers in the plurality of servers;
• and
• performing an autoscale operation to adjust a number of servers in the plurality of servers to provide better performance for the distributed service in view of the predicted load.