2026-01 Monthly summary for dbt-labs/metricflow focusing on performance optimization delivered in this period. Implemented instance-level memoization in the resolve_available_items function to cache results by (resolution_node_id, pattern_ids), dramatically reducing repeated expensive DAG traversals during query resolution. Impact highlights: - Key feature delivered: Memoization to resolve_available_items, enabling faster resolution in queries with large metric sets (e.g., 260+ metrics in metadata discovery scenarios). - Major performance improvement: Replaced repeated expensive DAG traversals with a cache, reducing per-call cost from O(N × DAG_traversal_time) to near O(1 × DAG_traversal_time) for repeated arguments. This mitigates CPU spikes and liveness probe risks observed in metricflow-server under heavy load. - Reliability and testing: Changes are instance-scoped and integrate with existing query resolution flow; existing tests pass. Committing reference: 51501432b2380b7a3296d3d9c7ed6e9864cc8e55 (Add memoization to resolve_available_items (#1964)).

November 2025 summary for dbt-labs/dbt-adapters: implemented stability improvements for the Snowflake connector, including a certificate lock timeout to prevent deadlocks and an upgrade to snowflake-connector-python 4.0.0. This work reduces production deadlock risk, enhances reliability of Snowflake operations, and demonstrates effective dependency management and collaboration.