The latest release of Db2, version 11.5, unleashed on the world in June 2019, is being marketed by IBM as “The AI Database.” AI, or artificial intelligence, has been increasing in popularity and utility for several years now. Indeed, AI technology is being infused into many aspects of our everyday lives, including in social media, health care, self-driving automobiles, virtual assistants (e.g., Siri, Alexa, etc.), and just about every business sector in some form. AI promises to be one of the biggest technological game-changers of this century as it enables computing devices and systems to take on more activities that heretofore had to rely on a human.
OK, so what does IBM mean by an AI database? That is a good question as AI can mean any number of things. Well, let’s start at the beginning: IBM Db2 is a relational database management system that delivers data management and analytics capabilities for transactional and data warehousing workloads. Db2 is built to provide high performance, actionable insights, data availability, and reliability. It includes many advanced features including in-memory technology (BLU Acceleration), advanced management and development tools, storage optimization, workload management, actionable compression, and continuous data availability with the pureScale option.
What is New?
The first new “thing” about Db2 11.5 is that IBM now includes it as part of their Hybrid Data Management Platform. The IBM Hybrid Data Management platform is built to enable enterprises to seamlessly access, share, and analyze data whether it is structured, semi-structured, or unstructured. Furthermore, the data can be easily accessed and analyzed, whether it is stored on premises, in a public cloud, in a private cloud, using open source, or in any combination of those deployments.
The movement to a hybrid data management capability is one that most modern DBMS offerings have undertaken. As more users want to deploy and take advantage of many different types and forms of data and on many different platforms and environments, a hybrid approach makes sense. Instead of supporting many different database systems and platforms from multiple vendors, a single, integrated hybrid platform such as IBM has delivered can proffer significant benefits in terms of training, support, usability, and even cost. But hybrid data management is not, in and of itself, an AI capability.
IBM Db2 11.5 also delivers new functionality to enable organizations to prepare their data management operations for machine learning. The capability of machine learning to stream, analyze, and learn from data sets without explicit programming can be extremely valuable, enabling users to develop and improve their models on the DBMS where the data resides.
And with hybrid data management you can empower users across your business with self-service access to data. Data scientists working with data can use the hybrid data management capabilities to access data relevant from various lines of business and from practically any location. The ability of users to self-serve access to data can improve development speed and reduce strain on the IT department. Of course, the users must have the requisite authority to access any data.
Another AI-related new feature is the Augmented Data Explorer, which delivers natural language querying capabilities. Instead of requiring rigid SQL syntax, developers instead can use a search engine-like interface to access Db2 data. And the results of the queries can be viewed as data visualizations or as summaries in natural language.
Finally, and perhaps most-intriguing to Db2 DBAs, the Machine Learning Query Optimizer adds AI and machine learning to Db2’s query optimization process. Michael Krafick, host of DiscoverDB2, has observed, "I'm truly excited to see the impact of a machine learning optimizer on Db2's existing cost-based optimizer. The thought that Db2 can improve an optimization path based off a learned experience is truly mind blowing."
Developers and DBAs have long struggled with query optimization but as machine learning processes get infused into query optimization, Db2 can learn what works best and optimize queries based on past experiences with your specific use cases, as well as best practices.
Additional New Features
Not all new features are AI-related, though. As with every release of Db2, there are enhancements that will appeal to different organizations for various reasons. Highlighted features include support for storage devices with 4K sectors, automatic recompression based on data growth, call level interface (CLI) driver enhancements, and improved firewall capabilities using a new registry variable.
Furthermore, there are many improvements to existing capabilities such as simplified installation and deployment, improved usability and adoption, large database enhancements, improved recovery time, SQL insert and update optimizations, new monitoring metrics, and improvements to BLU Acceleration.
Perhaps my favorite new feature is the streamlining and simplification of the number of Db2 editions which makes it easier to work with Db2. Now there are only three editions: Db2 Community, Db2 Standard, and Db2 Advanced. And moving from one to another does not require an additional download.
The Bottom Line
There is a lot to digest with this latest Db2 announcement from IBM. Although much of the fanfare has been associated with the new AI capabilities, there is much more going on here. Inclusion of Db2 in IBM’s new hybrid data management offering is significant, and the laundry list of new capabilities is great, too.