The Green Cloud Scheduler proactively detects the over provisioned computing resources and identifies the most appropriate adaptation decisions to dynamically adjust them to the incoming workload so that the data center overall energy consumption is minimized. The Green Cloud Scheduler may generate adaptation action plans consisting of consolidation actions (virtual task deployment and virtual task migration) and dynamic power management actions (set server to hibernate state and wake up server).

The adaptation decision process uses a MAPE (Monitoring, Analysis, Planning and Execution) based methodology.

In the monitoring phase, the information regarding data centre’s current state is collected and represented in a programmatic manner by means of an ontology based model called Energy Aware Context Model (more details in http://www.springerlink.com/content/8785323445771154/).

In the analysis phase, the data center greenness level is determined by evaluating a set of predefined Green Performance Indicators (GPIs) describing the data center servers’ optimal loading values.

In the planning phase, the sequence of adaptation actions to be executed for bringing the data center in a green state is determined, either by identifying similar previous situations which require the same adaptation actions or by using a reinforcement learning (what-if analysis) based approach (more details in http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=6108269).

In the execution phase the sequence of adaptation actions determined in the planning phase are enforced.

The Green Cloud Scheduler manages the data center's computing resources and the hosted virtual machines in an energy efficient way. The Green Cloud Scheduler increases the data center greenness by assuring that an optimal number of servers are used to accommodate and run the incoming workload.