The Next-Generation Towards Optimized Analytics

Teradata Collaborative Optimizer

The Collaborative Optimizer is applicable to both remote analytical functions, e.g., provided by MLE and TensorFlow engines, and native analytical functions built in the NewSQL engine. In this video, we will give an overview on the Collaborative Optimizer feature and show the significant performance gain it can provide to Teradata customers.

Collaborative Optimizer is the backend infrastructure of Teradata Vantage ecosystem for optimizing analytical functions and enabling analytics at scale. The Collaborative Optimizer is applicable to both remote analytical functions, e.g., provided by MLE and TensorFlow engines, and native analytical functions built in the NewSQL engine. The Collaborative Optimizer provides a declarative metadata-based mechanism, referred to as 'function descriptors', for capturing important properties of the analytical functions, through which various optimizations become feasible including: (1) Predicate Push, which entails pushing post-function predicates to the function's input for early elimination of unneeded rows, (2) Projection Push, which entails the elimination of unnecessary columns from the function's input, among other optimizations. In this session, we will give an overview on the Collaborative Optimizer feature and show the significant performance gain it can provide to Teradata customers.