본문으로 건너뛰기
다이제스트

Databricks 세계에서 무엇이 주목받았나.

릴리스, 뉴스, 동영상, 프로젝트, 커뮤니티 Q&A를 아우르는 하나의 이야기.

150 items · 4 themes · Nov 21, 10:51 UTC

각 주제 아래에 표시된 출처를 바탕으로 AI가 작성했습니다. 오류가 있을 수 있습니다. 오류 신고하기

Oct 22 to Nov 21, 2026

The period was dominated by Databricks positioning Lakebase as an operational engine with same-day PostgreSQL 19 support, alongside a broad expansion of Genie into autonomous agent execution and migration. Concurrently, new Unity Gateway tooling focused on agent governance, while platform SDKs and CLI releases standardized around Statement Execution V2.

1.Databricks launches Postgres 19 on Lakebase alongside automated operational diagnostics

Databricks announced same-day availability for PostgreSQL 19 on Lakebase, highlighting features like online space reclamation via REPACK and read-your-writes replication. The release drew active discussion on Hacker News regarding operational vacuum overhead, while Databricks released details on its caching architecture and introduced Lakebase Insights, an agent designed to flag deadlocks, slow queries, and missing indexes.

2.Genie expands beyond BI into autonomous pipeline repairs, code tasks, and serverless migration

Databricks expanded its Genie portfolio from text-to-SQL analytics into autonomous execution tooling. The vendor announced the public beta of Genie ZeroOps to remediate broken data pipelines using automated pull requests, launched Genie Migrate to translate legacy code and classic clusters to serverless infrastructure, and enabled autonomous Genie Code execution directly within Lakeflow jobs.

3.Unity Gateway introduces observability tooling to monitor agent token costs and MCP tool errors

Databricks launched Unity Gateway Insights to audit AI agent workloads, claiming it can automatically flag token waste caused by recurring Model Context Protocol (MCP) errors and routine prompts routed to expensive frontier models. Alongside benchmarking harnesses comparing Claude Code, Codex, and Hermes, Databricks released its AI Index to publish enterprise model usage trends derived from Unity Gateway telemetry.

4.Developer tooling advances with Statement Execution V2, PyPI CLI packaging, and Asset Bundle isolation

Databricks shipped coordinated updates across its developer ecosystem, making the Databricks CLI installable directly via PyPI with new sandbox commands. Simultaneously, the Databricks VS Code extension added catalog isolation mode for Databricks Asset Bundles, while the Python, Java, and Go SDKs introduced the Statement Execution V2 workspace service and security scan management endpoints for Databricks Apps.