Modern Cloud Analytics

This is a relatively abstract, but useful model. On the left, you have the assortment of sources that the modern enterprise has to deal with. These are databases, applications, streams and files.

In our experience, it makes the most sense to work towards the goal of storing these sources in a cloud data warehouse technology if the goal is doing analytical work. Generally, the best way to load these systems is to stage data in the object store of whatever cloud you’re in and then to either use the native load capability of the warehouse, or if you have a more complicated use case to utilize a data integration technology to schedule and orchestrate loads.

One thing worth pointing out is that modern cloud data warehouses are all very scalable and makes an ELT rather than an ETL pattern an attractive option to utilize this computing power to do our transformations in database.

 On the right side of the diagram are all of the usual downstream use cases that our clients present to us.