Impala and Big Query - discussion about dremel based technologies.

As we know today, MapReduce is a main workhorse for the large analytical data processing. And we also aware of its main drawbacks is a lack of interactiveness.

A few years ago Google published academic paper about the Dremel - technology used by google for interactive analysis of big data sets http://static.googleusercontent.com/external_content/untrusted_dlcp/research.google.com/en//pubs/archive/36632.pdf).
This paper is a basis for the several developed products: Impala, BigQuery, as well as several projects in progress:
Drill and OpenDremel.

Impala is an open source implementation of the Dremel concepts from Cloudera.
Big Query is a google's service based on Dremel.

Main topics to be discussed:
a) What is Impala, its main features and limitations. How it is integrated with Hadoop.
There will be some low-level technical details of the Impala implementation available from its sources.
b) Google Big Query. We will analyze its features and try to put on the table pros and cons of its usage.

I think this discussion will be interesting for people considering technologies to implement in order to make large data analysis more interactive.

I will be glad to get related topics and will do my best to prepare answers, if you can share those with me.
Your experience in these technologies - I will be glad to summarize and share it or give a slot for presentation.

We also will take case of what MediaMind are doing with big data and analyze what part of their workload can be optimized by using Impala.