|Data centers are ever increasing as we become more reliant on web based transactions. The benefits of such massive computing are obvious by the speed and ease we can get most media or information. A challenge is that new large data centers introduce a level of energy consumption that the world has not seen before. The obvious energy cost of running the computers is a billion dollar problem, but there are hidden costs like running cooling systems as well. Moreover, data centers are getting more concentrated on specific tasks, be it SQL or Hadoop or anything else an organization needs. To help combat the problems of large data centers, we aim at developing solutions that can work for each type of data center. This could entail creating tools that are generic enough to work for all data centers, or focusing on specific tools the type of software running in the data center. Our dissertation study works in both ways. We build a thermal model that is flexible enough to be applicable for all data centers; within our thermal model research we even show how it can be used to save energy. We also create energy saving techniques for Hadoop clusters specifically, where we focus on very data centric implementations of Hadoop to gain a significant energy savings. Lastly we propose a Spark specific process that takes what we have learned from Hadoop and thermal research and developed techniques that offer large energy and thermal savings within Spark clusters.