Projects

The project should be carried out individually but preferably in a group of two students. Each students should have complete knowledge of the system and be able to do the necessary experiments. The written presentations are done individually.

We might have more than one group assigned to a project. In that case the two groups will work independently.

When you have a good understanding of the goal of your own project, how to design a solution and which tools to use, you will give a presentation to students of other groups. You will also listen to their descriptions thus learning more about problems that you will face when implementing a distributed system.

A large part of you project will consist of performing performance studies, experiments that will help you in your design decisions. You will during the course have meetings with other students where you can present your findings and which decisions you have taken.

Towards the end of the project you should have compiled evaluations that convincingly show that you have met the goals of the project.

Projects

• Quota-based Scheduling for YARN
• True Snapshots for HDFS with BtrFS
• Egier a Geo Replicated System
• Partitioncentric Iterations in Apache Flink
• Performance Interference Characterization of Big Data Workloads

Deadeline

Papers should be handed in by mail no later than January 12'th.