SQL Server, Microsoft's flagship operational database, has been around for over a quarter century now. That has meant successive waves of innovation for the platform, but has also made for entrenchment of older versions, and accompanying challenges as periods of mainstream and extended support end. Tomorrow, another such rite of passage will take place as SQL Server's 2008 and 2008R2 releases exit extended support. But the Azure cloud and a forthcoming public preview of SQL Server 2019 Big Data Clusters on July 24th, should provide some consolation.
Also read: The big data odyssey of SQL Server 2019, and more data and AI news from Microsoft Ignite
SQL Server 2008/2008 R2 were the immediate successors to the 2005 release, which was a major reboot of the platform, parting ways with the original code base. By the time 2008R2 was released in 2010 (the same year original SQL Server collaborator Sybase was acquired by SAP), the product included components for self-service reporting; extract, transform and load (ETL); and master data management. 2008R2 also introduced Power Pivot, a precursor to today's Power BI. In fact, 2008R2 was known internally at Microsoft as "the BI refresh."
Approximately a decade later, it's time for these versions to be retired from extended support. Microsoft is using the occasion to encourage customers to move to the cloud. In a phone interview with ZDNet, Microsoft's Director, Operational Databases and Blockchain Product Marketing, Wisam Hirzalla, explained that Azure SQL Database ("SQL DB") is the compatible, cloud implementation of SQL Server. While several SKUs of SQL DB running in multi-tenant environments exist, Microsoft makes various specialized versions available as well.
For example, Azure SQL Database managed instance provides what Microsoft calls "evergreen SQL" -- in other words, an instance of the technology that is continuously updated and upgraded, and that is fully managed by the company. It also runs in an isolated virtual network and, according to Hirzalla, employs artificial intelligence/machine learning algorithms for tuning, threat detection and other self-management tasks. Meanwhile, SQL DB Hyperscale, general availability (GA) of which was announced in May at Microsoft's Build conference in Seattle, provides a great migration target for SQL Server customers with databases in the double-digit terabyte (TB) range.
Also read: Microsoft 'Builds' its data story, in the cloud and at the edge
Microsoft customers with Software Assurance can use the so-called "Azure Hybrid Benefit" to move their SQL Server deployments to Azure virtual machines (VMs), providing the configurability and flexibility associated with on-premises implementations, without incurring additional licensing expenses. While charges for the virtual machine itself and, in certain cases, data egress will apply, those replace whatever ongoing costs customers would have to purchase, house, power, cool, manage and maintain their own hardware. SQL on Azure virtual machine implementations also get three free years of extended security updates.
Microsoft says it will soon launch a portal experience, branded simply (if not confusingly) as "Azure SQL," that will allow customers to manage both Azure SQL Database and SQL Server on Azure VM implementations in a single environment. Azure SQL will also provide access to auto-patching, auto-backup and new license management options, providing Platform as a Service (PaaS)-like benefits to Infrastructure as a Service (IaaS) SQL Server implementations.
Microsoft has released four versions of SQL Server (2012, 2014, 2016 and 2017) subsequent to 2008R2, adding support for in-memory tables, columnar storage and machine learning as well as operation on Linux and in Docker containers. The next on-premises release of the platform, dubbed SQL Server 2019, adds integration of Apache Spark and the Hadoop Distributed File System (HDFS) through a feature called Big Data Clusters (BDC). SQL Server 2019 BDC has been in private preview for months, but the company is announcing that a public preview of the technology will begin in just a little over two weeks, on July 24th.
That's exciting, but one might ask why Microsoft now sees fit to integrate Big Data technology into SQL Server, over a decade after Hadoop first broke on the tech scene, and when an increasing number of data lakes are being implemented in the cloud. I posed these questions to Hirzalla, who provided what I found to be reasonable and nuanced responses.
To begin with, the SQL Server team sees Big Data as a reality but one sequestered from operational database and BI implementations. To paraphrase the company's talking points, that isolates the high-volume data, often managed by Hadoop and Spark, from the high-value data often managed by relational database and Enterprise BI platforms.
SQL Server 2019 BDC seeks to integrate these two regions of the enterprise "data estate," allowing each to leverage the other. The co-location of SQL Server instances and Spark worker nodes will allow both engines to operate efficiently on the very same data, stored in Parquet and other open data formats. And SQL Server 2019's enhanced PolyBase technology will allow the engine to connect to high-volume data and blend it with SQL Server-based transactional data and data marts.
Also read: Microsoft's PolyBase mashes up SQL Server and Hadoop
When it comes to cloud versus on-premises, the lines are similarly blurred. That's because SQL Server 2019 BDC builds on the Linux container-compatibility introduced in SQL Server 2017, with deployment based completely on Kubernetes (aka K8s). That, in turn, will make for easy deployment of the new version of SQL Server to both on-premises K8s clusters as well cloud environments including, but not limited, to Azure Kubernetes Service (AKS). Installation of Spark and HDFS comes along for the ride, greatly simplifying deployment of those technologies, too. As I said, these are reasonable answers to a couple of very fair questions.
One of the more impressive things about SQL Server is that it maintains compatibility with timeless database development concepts and patterns introduced decades ago, while also integrating new technologies and making them consumable by SQL Server professionals, on their own terms. In order to do that, though, the company does need its installed base to move off old versions.
That time has come for customers using SQL Server 2008 and 2008R2. For those customers, Microsoft has laid out a path forward with many lanes, some of them leading upward to the cloud.