1 year ago

Materialized views evoke cached results — stale data, which is antithetical to “real time.” Let’s take a look at how recent advancements in vectorized computation have flipped this mode of thinking to implement continuously updated materialized views that always show up-to-date results in response to real-time data feeds, even with complex queries.

Most databases deploy materialized views on relatively static data because of the computational cost of executing and refreshing complex queries on real-time data feeds.

Continuously updated materialized views: Continuously updated MVs can simplify data engineering by allowing developers to easily deploy end-to-end analytical pipelines that take in real-time data feeds and cascade complex computations through a series of materialized views to deliver fresh insights and results to downstream applications.

A continuously updated materialized view that handles real-time data feeds looks exactly the same as any other materialized view.