Skip to content

Learnt from previous projects

Seppala edited this page Aug 16, 2011 · 1 revision

We're starting to collect the nuggets of learning that the teams have uncovered while doing Hack/Reduce projects. The idea is that participants will be able to add the things they've learnt here. This is not meant to be a book about all the stuff, but just quick things like "Mandy works really well for Ruby, but Wukong doesn't", "When using streaming, don't use the same name for the mapper and reducer files, it will screw things up".

Streaming

Don't use the same names for the mapper and reducer files.

Ruby

One team tried Mandy and Wukong with Ruby. Mandy was easy to get to work, but they couldn't get Wukong to work. Use Mandy. You can also just directly use streaming, but Mandy should make it easier to use Hadoop.