Overview
This article details the improvements included in the dentsu.Composable Translate 2.1.0 release.
Improvements and updates
Asynchronous job execution for match requests
dentsu.Composable Translate now supports asynchronous execution for all _to_table match templates through APP_PUBLIC.RUN_ASYNC. Requests are submitted immediately while processing continues independently using the consumer's Snowflake warehouse, eliminating the need to keep sessions active while jobs run. The enhancement supports unlimited concurrent processing through dynamically created tasks, enables workload distribution across multiple warehouses, and provides the ability to cancel in-flight jobs when needed. This update improves scalability, reduces processing bottlenecks, and provides greater flexibility for managing large matching workloads.
Enhanced request status and job tracking details
The REQUEST_STATUS and REQUEST_OPTIONS views now include additional metadata to improve request tracking and troubleshooting. REQUEST_STATUS now includes KB_BUILD_ID, while REQUEST_OPTIONS now includes DENTSU_CLIENT_KEY and OUTPUT_LOCATION. These additions help users quickly identify the build used for processing, associate requests with specific clients, and locate generated outputs more efficiently.
Consistent matching logic for waterfall processing
Waterfall and waterfall_internal matching now enforce a defined match-priority hierarchy of HMID, clear email, hashed email, device ID, phone, and IP address. When multiple identifiers match the same record, the highest-priority identifier is selected automatically. This enhancement delivers more consistent and predictable match results across waterfall-based workflows.
Improved match performance through build-scoped caching
All match templates now use build-scoped cache tables keyed by the target Dentsu client key, including clear email, hashed email, device ID, IP address, phone, HMID, client CMID, waterfall, and waterfall_internal. While the first request after a new build may require additional processing time to create the cache, subsequent requests complete significantly faster. This enhancement improves overall performance, reduces processing time, and lowers compute costs for repeated match requests.
Improved administration, monitoring, and platform reliability
Several maintenance and usability enhancements have been implemented across the platform:
DEACTIVATE_INSTANCEnow returns a clear message when an invalid or inactive activation code is supplied.- Telemetry events are generated for activation and deactivation activities.
- Benchmarking processes now leverage parallel execution to reduce runtime and associated compute costs.
These improvements enhance platform observability, administrative transparency, and overall operational efficiency.