This is a great video that includes a a discussion about a technique for efficient, fault-tolerant pipelined parallel query processing. 

Also, make sure to sign up for the free Azure Cloud sample, no credit card required.

This video gives a few answers to questions around how do you simplify data management in large data analysis.  How do you implement efficient support for complex analytics.

Although the video overall is good, consider downloading the video and speeding it up.

The video reveals a better solution than the one in the slides (with the video you might need to refresh):

image

image