跳到正文
Databricks Blog·· 6 小时前AI 评分22

Azure Government 版 Azure Databricks 进入 Beta

The new Azure Databricks on Azure Government is now in Beta

AI 导读

Azure Databricks on Azure Government 已进入 Beta,采用与商业版 Azure Databricks 相同的架构,首批提供 Unity Catalog、Databricks SQL 和无服务器计算。

正文 · 原文

Modern data and AI, built for the mission

Government agencies and their mission-critical partners face a persistent dilemma: adopt the latest data and AI innovations, or meet the stringent security and compliance requirements their missions demand. For too long, these have been competing priorities, forcing public sector organizations to operate on platforms that lag behind their commercial counterparts, sometimes by years.

Today, we're excited to announce that Azure Databricks on Azure Government is entering Beta, a major milestone in our commitment to delivering the full power of the Databricks Data + AI Platform to the U.S. public sector. Government agencies and the hundreds of organizations that support them now have access to the modern data foundation already transforming industries across the commercial sector, including Unity Catalog, Databricks SQL, and serverless compute, with the complete suite of AI capabilities to follow, all within an environment working toward FedRAMP Class D (formerly “High”) and DoW IL5 authorization.

One Databricks platform, wherever your mission runs on Azure

Databricks recently launched the FedRAMP Class D offering on Azure Commercial. And today we are bringing the best of Databricks to Azure Government, so U.S. public sector customers now have a modern Databricks platform on both Azure clouds.

For federal civilian agencies, state and local governments, and partners whose workloads fit FedRAMP Class D, Azure Commercial is available today. For organizations that need a dedicated government cloud and a path to FedRAMP Class D and DoW IL5, the new Azure Government environment is now open for Beta.

As a first-party Azure service, Azure Databricks is managed through the Azure portal, integrates natively with Microsoft Entra ID, and counts toward existing Azure commitments, so agencies can adopt it inside the cloud, contracts, and identity systems they already trust.

What's available in the new environment

Unlike the legacy environment, the new Azure Databricks on Azure Government runs on the same architecture as commercial Azure Databricks. It is designed to receive the full Databricks platform, with every commercial feature coming to Azure Government through the same ongoing releases. Beta is the first release on that path, and it already gives government teams the core capabilities their commercial peers rely on.

  • One place to govern data and AI. For the first time on Azure Government, Unity Catalog lets teams manage access, track lineage, and audit usage across all data and AI assets. Define a policy once, apply it everywhere, and easily show who accessed what, when, and why.
  • Analytics without infrastructure overhead. Databricks SQL delivers fast, high-concurrency analytics. Engineers and data scientists can choose classic clusters or serverless compute for each workload, which means less time sizing and patching clusters and more time delivering for the mission.
  • Secure by default. Every workspace comes with the Compliance Security Profile already enabled. It provides enhanced monitoring, automatic cluster updates, a hardened compute image, and encrypted traffic between cluster nodes. Azure Private Link is required on the front end and back end, so all traffic stays on the Azure backbone.
  • Full visibility into usage and spend. System tables give administrators one queryable view of audit logs, billing, compute, pipelines, ML experiments, query history, and storage, so they can answer audit requests and manage costs with familiar tools.
  • Data stays in government regions. Customer content is stored and processed only in U.S. government cloud regions: US Gov Virginia and US Gov Arizona.

What's next: bringing the full power of AI to the mission

Beta lays the governed, secure foundation that AI depends on. Next, we'll bring the full Databricks AI suite to Azure Government, all governed by Unity Catalog, so every model, agent, and answer inherits the same access controls, lineage, and auditing as the underlying data. On the roadmap:

  • Conversational analytics with Genie. Ask questions in plain language and get accurate, governed answers, with no SQL and no waiting on a reporting queue.
  • AI-assisted development. Built-in AI helps engineers, analysts, and data scientists write, explain, and debug code faster.
  • Production-grade agents with Agent Bricks. Build, evaluate, and deploy agents that reason over your own documents and data, for example to search case files, summarize reports, or automate routine workflows. Built-in quality evaluation means teams can trust the results before putting them in front of users.
  • Models and serving. Model Serving and Foundation Model APIs let teams deploy their own models and use curated foundation models through a single governed endpoint.
  • AI on all your data. Vector Search and AI Functions power retrieval-augmented apps and let teams classify, extract, and summarize data directly in SQL, without moving it.
  • Data and AI applications. Databricks Apps and Lakebase let teams build secure, interactive applications and operational workloads directly on the lakehouse.

Roadmap capabilities and timelines are forward-looking and subject to change.

FedRAMP Class D and IL5 authorization

Databricks is actively pursuing FedRAMP Class D (formerly FedRAMP High) and DoW Impact Level 5 (IL5) authorizations for the new Azure Government environment, expected in early 2027. Once authorized, agencies and defense partners will be able to run Controlled Unclassified Information (CUI) and mission-critical workloads on the modern Databricks platform, including the defense workloads that require IL5 and are not supported on Azure Commercial.

Track our authorization progress on the FedRAMP Marketplace.

Getting started with Beta

New to the environment? Opt in with a dedicated Azure Government subscription that has never been used for Azure Databricks. After you register the new-environment preview feature in the Azure portal, every new workspace in that subscription is created in the new environment. During Beta, the environment is intended for non-production workloads. The workspaces you build now will carry forward to production at General Availability.

Already on the legacy Azure Government environment? Now is the time to start planning your migration. Your existing workspaces, notebooks, data, and configurations are unaffected, so you can explore the new environment side by side and work out your path forward. We recommend starting with four areas:

  1. Unity Catalog. Legacy Azure Government workspaces have never had Unity Catalog. Starting early gives your teams time to design and adopt unified governance.
  2. Serverless compute. Serverless is new to government environments, so begin your internal security and compliance review now.
  3. Network encryption. The Compliance Security Profile is enabled on every workspace and requires instance types that support Azure Virtual Network encryption. Update your cluster configurations and Terraform/ARM templates to match.
  4. Your ATO. Start working with your security and authorizing officials now to add Databricks on Azure Government to your system's Authorization to Operate. Updating system security plans, data flows, and boundary documentation during Beta puts you in a position to run CUI production workloads at General Availability, expected in early 2027.

Our Federal team can help you assess your workloads and plan the move.

Ready to get started? Contact your Databricks representative or check out the Azure Databricks documentation for technical details.

来源:Databricks Blog · databricks.com