Data-Intensive Application Architectures

Question 1

Downsides of the traditional development process

Answer 1

1 - The business value is lost throughout the process
2 - Limited flexibility
3 - Cost of failure is high
4 - Takes a lot of time

Question 2

Benefits of the big data process

Answer 2

1 - Iterative rapid cycles
2 - Lots of testing
3 - More innovative

Question 3

Lambda architecture

Answer 3

A reference architecture for big data systems and defined as a system that runs arbitrary functions on arbitrary data

Question 4

The three lambda layers

Answer 4

1 - Batch
2 - Speed
3 - Serving

Question 5

Lambda: batch layer

Answer 5

Contains the immutable, constantly growinng dataset

Question 6

Lambda: speed layer

Answer 6

Deals with new data and compensates for the high latency updates of the serving layer

Question 7

Lambda: serving layer

Answer 7

Loads and exposes the combined view of data such that they can be queried

Question 8

If you have a regular ML application, how do you change it such that you use the Lambda architecture?

Answer 8

Regular ML models only use the batch layer. Expand it by adding an interaction between the batch-created model and the streams of data

Question 9

Kappa Architecture

Answer 9

Everything is done in the streaming (speed) layer. The versioning of the application should be taken into account