Skip to content

polybase-guide.md

Latest commit

 

History

History
167 lines (117 loc) · 13.6 KB

polybase-guide.md

File metadata and controls

167 lines (117 loc) · 13.6 KB
title Introducing data virtualization with PolyBase
description PolyBase enables your SQL Server instance to process Transact-SQL queries that read data from external data sources such as Azure Blob Storage.
ms.date 08/22/2022
ms.service sql
ms.subservice polybase
ms.topic overview
f1_keywords
PolyBase
PolyBase, guide
helpviewer_keywords
PolyBase
PolyBase, overview
Hadoop import
Hadoop export
Hadoop export, PolyBase overview
Hadoop import, PolyBase overview
ms.custom
contperf-fy21q2
intro-overview
event-tier1-build-2022
author WilliamDAssafMSFT
ms.author wiassaf
ms.reviewer hudequei
monikerRange >=sql-server-2016||>=sql-server-linux-ver15||>=aps-pdw-2016||=azure-sqldw-latest

Introducing data virtualization with PolyBase

[!INCLUDEappliesto-ss-xxxx-asdw-pdw-md]

PolyBase is a data virtualization feature for [!INCLUDEssNoVersion].

What is PolyBase?

PolyBase enables your [!INCLUDEssNoVersion] instance to query data with T-SQL directly from [!INCLUDEssNoVersion], Oracle, Teradata, MongoDB, Hadoop clusters, Cosmos DB, and S3-compatible object storage without separately installing client connection software. You can also use the generic ODBC connector to connect to additional providers using third-party ODBC drivers. PolyBase allows T-SQL queries to join the data from external sources to relational tables in an instance of [!INCLUDEssNoVersion].

A key use case for data virtualization with the PolyBase feature is to allow the data to stay in its original location and format. You can virtualize the external data through the [!INCLUDEssNoVersion] instance, so that it can be queried in place like any other table in [!INCLUDEssNoVersion]. This process minimizes the need for ETL processes for data movement. This data virtualization scenario is possible with the use of PolyBase connectors.

Supported SQL products and services

PolyBase provides these same functionalities for the following SQL products from Microsoft:

Note

Data virtualization using PolyBase feature is available in preview for Azure SQL Managed Instance, scoped to querying external data stored in files in Azure Data Lake Storage (ADLS) Gen2 and Azure Blob Storage. Visit Data virtualization with Azure SQL Managed Instance to learn more.

SQL Server 2022 PolyBase enhancements

New to [!INCLUDEsssql22-md] Details
S3-compatible object storage [!INCLUDEsssql22-md] adds new connector, S3-compatible object storage, using the S3 REST API. You can use both OPENROWSET and EXTERNAL TABLES to query data files in S3-compatible object storage.
Some connectors separate from PolyBase services The S3-compatible object storage connector, as well as ADSL Gen2, and Azure Blob Storage, are no longer dependent of PolyBase services. PolyBase services must still run to support connectivity with Oracle, Teradata, MongoDB, and Generic ODBC. The PolyBase feature must still be installed on your SQL Server instance.
Parquet file format PolyBase is now capable of querying data from Parquet files stored on S3-compatible object storage. For more information, see to Virtualize parquet file in a S3-compatible object storage with PolyBase.
Delta table format PolyBase is now capable of querying (read-only) data from Delta Table format stored on S3-compatible object storage, Azure Storage Account V2, and Azure Data Lake Storage Gen2. For more information, see to Virtualize Delta Table format
Create External Table as Select (CETAS) PolyBase can now use CETAS to create an external table and then export, in parallel, the result of a [!INCLUDEtsql] SELECT statement to Azure Data Lake Storage Gen2, Azure Storage Account V2, and S3-compatible object storage. For more information, see CREATE EXTERNAL TABLE AS SELECT (Transact-SQL).

For more new features of [!INCLUDEsssql22-md], see What's new in SQL Server 2022?

For an example using PolyBase in [!INCLUDEsssql22-md] to virtualize a CSV file in Azure Storage, see Virtualize CSV file with PolyBase.

PolyBase connectors

The PolyBase feature provides connectivity to the following external data sources:

External data sources [!INCLUDEssNoVersion] 2016-2019 with PolyBase [!INCLUDEsssql22-md] with PolyBase APS PDW [!INCLUDEssazuresynapse_md]
Oracle, MongoDB, Teradata Read Read No No
Generic ODBC Read (Windows Only) Read (Windows Only) No No
Azure Storage Read/Write Read/Write Read/Write Read/Write
Hadoop Read/Write No Read/Write No
[!INCLUDEssNoVersion] Read Read No No
S3-compatible object storage No Read/Write No No
  • [!INCLUDEsssql22-md] does not support Hadoop.
  • [!INCLUDEsssql16-md] introduced PolyBase with support for connections to Hadoop and Azure Blob Storage.
  • [!INCLUDEsssql19-md] introduced additional connectors, including [!INCLUDEssNoVersion], Oracle, Teradata, and MongoDB.
  • [!INCLUDEsssql22-md] introduced the S3-compatible storage connector.

Examples of external connectors include:

* PolyBase supports two Hadoop providers, Hortonworks Data Platform (HDP) and Cloudera Distributed Hadoop (CDH), through SQL Server 2019. [!INCLUDEpolybase-java-connector-banner-retirement]

To use PolyBase in an instance of [!INCLUDEssNoVersion]:

  1. Install PolyBase on Windows or Install PolyBase on Linux.
  2. Starting with [!INCLUDEsssql19-md], enable PolyBase in sp_configure, if necessary.
  3. Create an external data source.
  4. Create an external table.

Azure integration

With the underlying help of PolyBase, T-SQL queries can also import and export data from Azure Blob Storage. Further, PolyBase enables [!INCLUDEssazuresynapse_md] to import and export data from Azure Data Lake Store, and from Azure Blob Storage.

Why use PolyBase?

PolyBase allows you to join data from a [!INCLUDEssNoVersion] instance with external data. Prior to PolyBase to join data to external data sources you could either:

  • Transfer half your data so that all the data was in one location.
  • Query both sources of data, then write custom query logic to join and integrate the data at the client level.

PolyBase allows you to simply use Transact-SQL to join the data.

PolyBase does not require you to install additional software to your Hadoop environment. You query external data by using the same T-SQL syntax used to query a database table. The support actions implemented by PolyBase all happen transparently. The query author does not need any knowledge about the external source.

PolyBase uses

PolyBase enables the following scenarios in [!INCLUDEssNoVersion]:

  • Query data stored in Azure Blob Storage. Azure Blob Storage is a convenient place to store data for use by Azure services. PolyBase makes it easy to access the data by using T-SQL.

  • Query data stored in Hadoop from a [!INCLUDEssNoVersion] instance or PDW. Users are storing data in cost-effective distributed and scalable systems, such as Hadoop. PolyBase makes it easy to query the data by using T-SQL.

  • Import data from Hadoop, Azure Blob Storage, or Azure Data Lake Store. Leverage the speed of Microsoft SQL's columnstore technology and analysis capabilities by importing data from Hadoop, Azure Blob Storage, or Azure Data Lake Store into relational tables. There is no need for a separate ETL or import tool.

  • Export data to Hadoop, Azure Blob Storage, or Azure Data Lake Store. Archive data to Hadoop, Azure Blob Storage, or Azure Data Lake Store to achieve cost-effective storage and keep it online for easy access.

  • Integrate with BI tools. Use PolyBase with Microsoft's business intelligence and analysis stack, or use any third-party tools that are compatible with [!INCLUDEssNoVersion].

Performance

  • Push computation to Hadoop. (Applies to [!INCLUDEsssql16-md], [!INCLUDEsssql17-md], and [!INCLUDEsssql19-md] only.) PolyBase pushes some computations to the external source to optimize the overall query. The query optimizer makes a cost-based decision to push computation to Hadoop, if that will improve query performance. The query optimizer uses statistics on external tables to make the cost-based decision. Pushing computation creates MapReduce jobs and leverages Hadoop's distributed computational resources. For more information, see Pushdown computations in PolyBase.

  • Scale compute resources. (Applies to [!INCLUDEsssql16-md], [!INCLUDEsssql17-md], and [!INCLUDEsssql19-md] only.) To improve query performance, you can use [!INCLUDEssNoVersion] PolyBase scale-out groups. This enables parallel data transfer between [!INCLUDEssNoVersion] instances and Hadoop nodes, and it adds compute resources for operating on the external data.

[!INCLUDEpolybase-scaleout-banner-retirement]

Upgrading to SQL Server 2022

Starting in [!INCLUDE sssql22-md] Hortonworks Data Platform (HDP) and Cloudera Distributed Hadoop (CDH) are no longer supported. Due to these changes, it is required to manually drop PolyBase external data sources created on previous versions of SQL Server that use TYPE = HADOOP or Azure Storage before migrating to [!INCLUDE sssql22-md]. Dropping external data sources also requires dropping the associated database objects, such as database scoped credentials and external tables.

Azure Storage connectors must be changed based on the reference table below:

External Data Source From To
Azure Blob Storage wasb[s] abs
ADLS Gen 2 abfs[s] adls

Next steps

Before using PolyBase, you must install PolyBase on Windows or install PolyBase on Linux, and enable PolyBase in sp_configure if necessary. For more tutorials on creating external data sources and external tables to a variety of data sources, see PolyBase Transact-SQL reference.

Review PolyBase Transact-SQL reference with examples of external data sources and external tables for a variety of data sources. For more tutorials, review the following articles: