With Tecton 0.9, we're excited to announce the Public Preview of Rift: Tecton’s AI-optimized, Python-native compute engine.
We've worked closely with Private Preview customers to enhance Rift's capabilities and make it ready for broader availability. Rift Public Preview includes powerful capabilities that make Tecton significantly more secure, unlock new features that increase model quality, improve performance and reliability, and much more.
But that’s not all - Tecton 0.9 also includes several other enhancements that make feature serving & storage significantly more performant, improve the interoperability between Tecton and your data stack, and enhance the developer experience.
To get started with Tecton 0.9, please refer to our upgrade guide. We look forward to your feedback!
Rift Public Preview
The Public Preview of Rift adds support for several Tecton capabilities that make Rift more powerful than ever. This includes extended support for Tecton's Aggregation Engine, such as Offset Windows, Secondary Key Aggregates, Approximate Percentile, and Approximate Count Distinct. This also includes Feature Tables, Incremental Backfills, Array, Map, and Struct Data Types, Tecton's new Offline Retrieval Methods, and enhanced Unit Testing support.
Tecton Secrets (Public Preview)
0.9 includes the launch of Tecton Secrets, a secure mechanism for configuring and using credentials within all parts of Tecton.
With Tecton Secrets, customers can trust that credentials are securely handled and retrieved for use in Tecton capabilities like Data Sources and feature definitions.
Tecton Secrets are launching first for Rift, with Spark support to follow shortly.
Python Environments for Rift Batch & Stream Feature Views
0.9 unlocks the ability for Rift Batch & Stream features to rely on Python dependencies (including arbitrary packages that can be installed from pip!). Now, users can easily define their own custom Python Environment via a requirements.txt file.
Data Compaction for Spark Batch & Stream Feature Views (Private Preview)
Tecton 0.9 adds Data Compaction for Spark-based Stream Feature Views with Lifetime Window Aggregations.
Data Compaction is a powerful capability that optimizes the online performance and storage of Aggregation features in Tecton. With Data Compaction enabled, customers will see significantly faster online read times, reduced online store reads, and more.
Publish Features to Warehouse (Public Preview)
With 0.9, Publish Features to Warehouse launches to Public Preview. This capabilitiy improves interoperability with users' data ecosystem.
Tecton will automatically compute feature values and make them available in customers’ data warehouse (e.g. Snowflake), enabling feature analysis, exploration, selection, and evaluation.
Enhanced Offline Retrieval Methods (General Availability)
0.9 also includes the General Availability of Tecton's enhanced Offline Retrieval Methods for retrieving offline feature values during development and productionization.
Materialization Runtime Versioning for Rift-based Feature Views
As with Materialization Runtime Versioning for Spark Feature Views in 0.8,
Tecton 0.9 includes
versioning of the materialization runtime
that executes for the materialization of Rift Feature Views. This is configured
using environment, a new required parameter in 0.9.0 for Rift-based Feature
Views.
Updates to Default Spark Runtimes
Tecton 0.9 upgrades the default Spark runtimes used for the materialization of Spark-based Feature Views. The default for Databricks is updated to 11.3 LTS and the default for EMR is updated to 6.9.1.
