Implementing Spark like system in Haskell

The session will present design and implementation of Hspark. A library that implements a framework to enable running a distributed map-reduce job over a set of nodes. The session will also showcase an extensible DSL to specify distributed map-reduce job.

The session will focus mainly on Creation of DSL (Specification) for map reduce. The DSL is similar (actually based on) Apache Spark Translation of DSL into scheduling the jobs across the nodes, and Executing and handling failures. Current implementation of hspark is at and implements first two points mentioned above. Currently, I am trying to enforce it with separation of execution framework so that failures can be handled correctly. Note that this project was implemented as a part of course project for CS240H at Stanford. See the implementation details.

Length: 01:17:39
Views 70 Likes: 0
Recorded on 2016-10-13 at Functional Conf
Look for other videos at Functional Conf.
Tweet this video