Skip to content

Latest commit

 

History

History
53 lines (39 loc) · 1.58 KB

node-cluster-as-really-big-computer.md

File metadata and controls

53 lines (39 loc) · 1.58 KB
Raw

#Programming a Node Cluster Like It's One Really Big Computer

##Abstract

Distributed programming is a lot easier if the abstraction makes the cluster look just like a really big computer.

I'll talk a little bit about some concepts from Cloud Haskell (and Erlang and Akka), and how they might look ported over to Node. The main idea is to be able to spawn many Processes which are automatically distributed across the cluster. Processes communicate via message passing, inspired by Erlang's "actor model".

One point of focus is serialization and transmission of function closures. For example, given a function add

var add = function(x) {
  return function(y) {
    return x + y + 1
  }
}

We should be able to partially apply it and transmit it elsewhere to have its second argument applied, then have the result returned back to us.

spawnRemote(add(1), function(err, result) {
  if (err) throw err
  console.log("Result computed remotely: ", result)
})

Serialization of a function closure is possible by serializing its environment (i.e. the bindings of its free variables) and a representation of its code.

##Speaker Bio

neuman

Sr Software Engineer at Twitter, formerly at Twilio, board member at HIFY. Ideas man, Haskeller, a cynic with a bleeding heart.