2020 年 2 月February 2020

这些功能和 Azure Databricks 平台改进已于 2020 年 2 月发布。These features and Azure Databricks platform improvements were released in February 2020.

备注

发布分阶段进行。Releases are staged. 在初始发布日期后,可能最长需要等待一周,你的 Azure Databricks 帐户才会更新。Your Azure Databricks account may not be updated until up to a week after the initial release date.

用于基因组学的 Databricks Runtime 6.4 正式版Databricks Runtime 6.4 for Genomics GA

2020 年 2 月 26 日February 26, 2020

用于基因组学的 Databricks Runtime 6.4 是在 Databricks Runtime 6.4 基础上构建的。Databricks Runtime 6.4 for Genomics is built on top of Databricks Runtime 6.4. 它包含用于基因组学的 Databricks Runtime 6.3 的许多改进和升级。It includes many improvements and upgrades from Databricks Runtime 6.3 for Genomics.

关键功能包括:The key features are:

  • 现可自定义 DNASeqYou can now customize DNASeq. 管道用户可选择性地禁用读取比对、变体识别和变体批注阶段的任意合法组合。Pipeline users can selectively disable any legitimate combination of the read alignment, variant calling, and variant annotation stages. 用户还可执行单端读取比对。Users can also perform single-end read alignment.
  • 用于基因组学的 Databricks Runtime 6.4 中包含的 Glow 版本现在为以前仅通过 SQL 表达式公开的函数提供了 Python 和 Scala API。The version of Glow included in Databricks Runtime 6.4 for Genomics now provides Python and Scala APIs for functions previously exposed only via SQL expressions. 这些函数可用于 DataFrame 操作,从而提高了编译时安全性。These functions are available for DataFrame operations, providing improved compile-time safety.

有关详细信息,请参阅用于基因组学的 Databricks Runtime 6.4 的完整发行说明。For details, see the complete Databricks Runtime 6.4 for Genomics release notes.

Databricks Runtime 6.4 ML 正式版Databricks Runtime 6.4 ML GA

2020 年 2 月 26 日February 26, 2020

Databricks Runtime 6.4 ML 正式版引入了库升级,其中包括:Databricks Runtime 6.4 ML GA brings library upgrades, including:

  • PyTorch:1.3.1 到 1.4.0PyTorch: 1.3.1 to 1.4.0
  • Horovod:0.18.2 到 1.19.0Horovod: 0.18.2 to 1.19.0

有关详细信息,请参阅完整的 Databricks Runtime 6.4 ML 发行说明。For details, see the complete Databricks Runtime 6.4 ML release notes.

Databricks Runtime 6.4 正式版Databricks Runtime 6.4 GA

2020 年 2 月 26 日February 26, 2020

Databricks Runtime 6.4 正式版引入了新功能、改进和许多 bug 修补程序。Databricks Runtime 6.4 GA brings new features, improvements, and many bug fixes.

  • 用自动加载程序(公共预览版)增量处理新的数据文件。Process new data files incrementally with Auto Loader (Public Preview). 当新的数据文件在 ETL 过程中到达云 Blob 存储时,你可使用自动加载程序更高效地以增量方式处理这些文件。Auto Loader gives you a more efficient way to process new data files incrementally as they arrive on a cloud blob store during ETL. 这是对基于文件的结构化流的改进,它通过列出云目录和跟踪已查看的文件来识别新文件,随着目录的扩大,它的效率可能会非常低。This is an improvement over file-based structured streaming, which identifies new files by repeatedly listing the cloud directory and tracking the files that have been seen, and can be very inefficient as the directory grows.
  • 将数据加载到具有幂等重试的 Delta Lake(公共预览版)。Load data into Delta Lake with idempotent retries (Public Preview). 通过 SQL 命令 COPY INTO,可将数据加载到具有幂等重试的 Delta Lake(公共预览版)。The COPY INTO SQL command lets you load data into Delta Lake with idempotent retries (Public Preview). 若要将数据加载到 Delta Lake,现在必须使用 Apache Spark DataFrame API。To load data into Delta Lake today you have to use Apache Spark DataFrame APIs. 如果在加载过程中出现故障,必须有效地处理它们。If there are failures during loads, you have to handle them effectively.
  • 对 Delta 表的所有写入、更新和删除操作的操作指标现显示在表历史记录中。Operation metrics for all writes, updates, and deletes on a Delta table now shown in table history.
  • Azure Databricks 笔记本(公共预览版)中默认启用内联 Matplotlib 图。Inline Matplotlib figures now enabled by default in Azure Databricks notebooks (Public Preview).

有关详细信息,请参阅完整的 Databricks Runtime 6.4 发行说明。For details, see the complete Databricks Runtime 6.4 release notes.

新的交互式图表提供丰富的客户端交互New interactive charts offer rich client-side interactions

2019 年 2 月 25 日 - 3 月 3 日:版本 3.14Feb 25 - March 3, 2019: Version 3.14

此版本引入了两种新的交互式图表类型,它们会取代条形图和折线图实现形式。This release introduces two new interactive chart types that replace the bar chart and line chart implementations. 除了现有的图表功能以外,折线图还具有几个新的自定义绘图选项:设置 Y 轴范围、显示/隐藏标记,以及将日志比例应用到 Y 轴。In addition to existing chart functionality, the line chart has a few new custom plot options: setting a Y-axis range, showing or hiding markers, and applying log scale to the Y-axis. 两种图表都具有内置工具栏,后者支持一组丰富的客户端交互。Both charts have a built-in toolbar that supports a rich set of client-side interactions.

图表工具栏Chart toolbar

如果要使用现有的图表实现形式,可从“旧版图表”下拉菜单中选择它们。If you want to use the existing chart implementations, you can select them from the Legacy Charts drop-down menu. 现有图表将继续使用之前可用的实现形式。Existing charts will continue to use the previously available implementations.

旧版图表类型Legacy chart types

新的数据引入网络添加了与 Delta Lake 的合作伙伴集成(公共预览版)New data ingestion network adds partner integrations with Delta Lake (Public Preview)

2020 年 2 月 24 日February 24, 2020

现在,你可轻松地将你的“lakehouse”从数百个数据源填充到 Delta Lake;其中 lakehouse 是你的数据库,依托于你通常借助数据仓库获取的各种数据结构和数据管理功能。Now you can easily populate your “lakehouse”—your data lake empowered by the kinds of data structures and data management features you typically get with a data warehouse—from hundreds of data sources into Delta Lake. 此网络的核心是新的合作伙伴集成库,可从你的工作区进行访问,还可借助它通过我们的合作伙伴 Fivetran、Qlik、Infoworks、Streamsets 和 Syncsort 访问大型数据源网络。At the heart of this network is the new Partner Integrations gallery, accessible from your workspace and providing access to a huge network of data sources via our partners Fivetran, Qlik, Infoworks, StreamSets, and Syncsort.

合作伙伴集成门户Partner integrations portal

有关概述,请参阅我们的博客For an overview, see our blog. 有关详细信息,请参阅合作伙伴数据集成For details, see Partner data integrations.

工作区创建者自动添加为 Azure Databricks 管理员Workspace creator automatically added as an Azure Databricks admin

2020 年 2 月 24 日February 24, 2020

在 2020 年 2 月 24 日之前,仅当创建了 Azure Databricks 工作区的用户还在 Azure 门户中单击了“启动工作区”按钮时,该用户才被添加为该工作区的管理员用户,或者该用户由工作区中已经是管理员用户的用户添加为管理员(订阅中任何单击了“启动工作区”按钮的 Azure 参与者都会被创建为该工作区中的管理员用户) 。Before February 24, 2020, the user who created an Azure Databricks workspace would only be added as an admin user for the workspace if she also clicked the Launch Workspace button in the Azure Portal or was added as an admin by a user who was already an admin user in the workspace (any Azure Contributor for the subscription who clicked the Launch Workspace button would be created as an admin user in the workspace). 而现在,创建工作区的用户将被自动添加为工作区管理员。Now the user who creates the workspace will be added automatically as a workspace admin.

若要详细了解如何创建和启动工作区,请参阅管理订阅For details about creating and launching workspaces, see Manage your subscription

用于管理工作区安全性和笔记本功能的标志现已可用Flags to manage workspace security and notebook features now available

2020 年 2 月 4 日至 11 日:版本 3.12February 4-11, 2020: Version 3.12

此版本引入了新的标记,它们用于管理发送来阻止攻击工作区的安全性标头,以及笔记本结果下载和 Git 版本控制的访问权限。This release introduces new flags for managing the security headers that are sent to prevent attacks on your workspace, as well as access to notebook results downloads and Git versioning. 请参阅管理工作区安全性标头管理对笔记本功能的访问权限See Manage workspace security headers and Manage access to notebook features. 所有这些管理选项均默认启用。All of these administrative options are enabled by default.