Transient and map reduce cloud frameworks: Spark (Apache) and MBrace (FSharp)

There is an article that details the similarities of Spark.

Transient has an implementation of distributed map-reduce over the basic Transient Cloud primitives: wormholeand teleport.

In transient the program define the architecture, while Spark and MBrace seems to be map-reduce engines which have a lot of extra developments over them for machine learning and data analysis. Although MBrace has more granularity and richer primitives. Transient is general purpose: it has a set of effects to be used for whatever kind of need. The same transient monad is used for creating web widgets inside web applications that may call distributed programs, or not.

The specialization of Spark and MBrace force them to rely on the need of additional libraries for doing other task like User interfaces, reactive programming or web programming that add extra complexity and do not enjoy functional composability. In transient it is possible to create full stack components that may work stand-alone or may be combined to create bigger applications. A distributed map-reduce operation can be part of a web widget that can be combined with others using functional composition.

To be completed