Skip to Content

You are currently visiting an old archive website with limited functionality. If you are looking für the current Berlin Buzzwords Website, please visit https://berlinbuzzwords.de

Urania Berlin, June 6-7, 2011

Stratosphere Hackathon

When
  • June 8/9/10

Where

  • Research group DIMA at TU Berlin on map

What to bring

  • Your Berlin Buzzwords badge to be admitted to the Hackathon.
  • Enthusiasm,
  • Creativity and expertise
  • Your laptop (or other hardware you need) preloaded with your favourite programming tools.

What

Stratosphere is an Open-Source system (https://www.stratosphere.eu), allowing you to do analytics beyond MapReduce. It features a programming model that generalizes MapReduce with multiple additional concepts, and a powerful parallel dataflow engine. The system solves many shortcomings of common MapReduce engines, like Hadoop, and gives in many cases a cleaner and more efficient approach to parallel data processing.

The system is highlighted by the following features:

  • Easy definition and massively parallel execution of complex data analysis tasks.
  • PACT is a generalization and extension of the well-known MapReduce programming model.
  • A cost-based optimizer compiles PACT programs to Nephele dataflow graphs.
  • Nephele executes dataflow graphs in a massively parallel and very flexible fashion.

With the hackathon, we aim both at users and at people who enjoy hacking the internals of a such a system (or those who want to get into it). The following are suggestions and ideas for what to do:

  1. Users: Learn to write PACT programs. Write your own tasks or port some existing MapReduce algorithms (e.g. Mahout, Matrix operations, ...) to PACT.
  2. Developers: Learn about the Stratosphere internals. Improve the system by adding new internal data processing strategies such as hash-based combiners, a memory-adaptive sort algorithm, or a new partitioning strategy.

During the Hackathon, food (lunch + snacks) and drinks (coffee, coffee, water, softdrinks, and juice) will be provided.

Registration

The hackathon will take place at TU Berlin, at June 8./9., plus one spare day, if we want more time. If you are interested, please send a mail to stephan.ewen@tu-berlin.de.

We are looking forward to all you great hackers signing up!