Automated tuning of MapReduce

Document Store, a distributed document storage solution developed by Yahoo! Technologies Norway. A working prototype of a MapReduce implementation has been developed using Vespa Document Store. However, this prototype is quite immature, and manual tuning of several parameters is required. Most of these parameters affect each other and the system as a whole in a complex manner, and substantial research is required to get a complete understanding of them. This thesis will focus on researching these parameters, so their effects can be fully known. Where applicable, automated tuning of parameters will also be researched. The research will consist of both theoretical modeling of effects and practical verification of results