Sadly the article http://venturebeat.com/2016/02/09/google-has-quietly-launched-its-answer-to-aws-lambda/ went slightly unnoticed as the original announcement made by Google. But what happenend?
Amazon is in production with its Lambda since a while, Microsoft is on its way too.
Why is it so important? The key point is that is an aknowledge of AWS approach to the problem: “Ok, let’s provide a similar service, dividing Big Data processing into two separate use cases: the batch -with cloud storage, cloud computing, cloud anything…- and the reactive one -queues, messages, events, IoT“. So it’s the subscription of Nathan Martz approach to Big Data: the so-called Lambda Architecture. Finally today I, being I a developer or a company, am able to develop an asynchronus event drive process without the burden of setting up an infrastructure. Clearly with some lock-in coming from cloud vendors. That’s why they continuosly change names to solutions… the one depicted today could have been Apache Storm or Akka being open source. Their ecosystem is branded.
There’s one noticeable side effect (we’ll come again on this) the simplicity of this model, not only compared to Map Reduce but also to Spark. The unlocking key is the “care about one event at a time” and this dramatically semplifies the handling of udates, which are the real trade off of (micro) batching approaches.
From this point you can already have the K-Architecture in sight…