• Replicated Search Engines
• Query Scheduling
• PESOS
• Network Simulator
• Energy Efficiency
• Distributed Systems
• CPU Energy Model

Web search engines continuously crawl and index an immense number of Web pages to return fresh and relevant results to the users’ queries. Web search engines operate on top of a distributed infrastructure composed by thousands of server nodes, hosted in large data-centers.
There are several factors contributing to the overall energy consumption of such data-centers. The main source of energy expenditure comes from the CPU. Hence, several works in the literature have proposed techniques to reduce the CPU energy consumption of search engines without degrading their performance.
In this thesis, we implement a simulator to evaluate the energy saving of the state-of-the-art techniques. More specifically, we evaluate the PESOS algorithm when deployed in a monolithic, distributed, and replicated search engine. In particular, we compare it with the industry-level PEGASUS algorithm, deployed by Google.
Our simulations show that, in a monolithic environment, PESOS can reduce its energy consumption up to ∼16% respect to its original version. The distributed version of PESOS outperforms PEGASUS by a ∼30% in energy consumption without incurring in a significant performance degradation. In the replicated environment we can reduce the energy consumption of PESOS by a further ∼2%.