Oracle Database 23ai: The Database Just Got a Brain

The database world is undergoing a renaissance. Fueled by the rise of Large Language Models, hybrid cloud data management, and the explosion of data-driven applications, databases are once again at the forefront of technological innovation.

Yet, this resurgence comes at a time of unprecedented challenges. The relentless onslaught of data and the ever-present threat of cyberattacks have stretched database professionals to their limits.

Oracle, a behemoth in the database landscape, has responded with a daring gambit: Oracle Database 23ai. This isn’t merely an update; it’s a full-blown integration of AI and machine learning directly within the database engine.

“We’re injecting AI into the database engine to make it easier and, in some ways, more bulletproof,” declares Jenny Tsai, Oracle’s senior vice president for overall database product management. Database made simple

Redefining simplicity through convergence

Oracle Database 23ai, previously born as 23c, is not your grandfather’s relational database. It’s a converged database that seamlessly handles various data types — JSON documents, graph data, and spatial data — all under one data-laden roof. This unified approach streamlines data architecture, eliminating the need for a patchwork of disparate database engines and the integration headaches that follow it.

Critical features like JSON Relational Duality empower developers to retrieve and store data using JSON effortlessly. At the same time, the Operational Property Graph enables them to build applications that navigate complex data relationships quickly. The result? Unprecedented choice and simplicity.

As Tsai points out, “The more different kinds of database engines you have to manage and try to integrate, the more complex your architecture is going to become.”

Vectors now have one home

GenAI has thrust vector databases into the limelight. These databases store and provide lightning-fast access to structured and unstructured data alongside vector embeddings, the numerical representations of data’s semantic meaning. They act as the external brain for many stateless LLMs, providing state, mitigating hallucinations, and underpinning Retrieval-Augmented Generation (RAG).

Oracle Database 23ai introduces a native VECTOR data type optimized for storing and searching vector data, which is “very easy for anybody [with a data science background] to select statements and do similarity search.”

You can even import ONNX-compatible embedding models directly into the database, generating vector embeddings on the fly using SQL, allowing you to create vector embeddings directly within the Oracle Database using SQL.

Powered and optimized by English

One of the most transformative aspects of Oracle Database 23ai is its natural language interface. Gone are the days of wrestling with complex SQL queries. Now, you can ask questions in plain English and get intelligent answers, thanks to the infusion of GenAI into the database engine.

But it’s not just about spitting out the right answers; it’s also about optimizing the entire database experience. AI-powered insights, automatic SQL tuning, and personalized recommendations for DBAs and developers remove the drudgery of manual tasks, freeing up time for strategic initiatives.

Tsai envisions a future where “tuning is no longer a job that the DBA has to do.” Instead, DBAs can focus on higher-value tasks like data stewardship and strategic planning.

Pssst, there’s a firewall inside

One striking feature of Oracle Database 23ai is the introduction of the In-Database SQL Firewall. SQL injection attacks have been a bane for DBAs for years, and Oracle’s response is to bring the fight directly to the data’s doorstep.

Tsai explains, “To really protect the data in the database, it’s best to have something sitting right inside the database.” This isn’t just a theoretical advantage; it’s a practical one. The in-database firewall is trained on legitimate SQL statements, allowing it to identify and block malicious queries accurately. It inspects, evaluates, and blocks database connections and SQL statements without the associated performance overhead.

This is not new. Oracle Database Firewall has been around for over a decade. But it was installed on the network to monitor SQL transactions. Now, it comes as part of the database itself (and its license), making Oracle Database 23ai secure once you spin it up.

Oracle Database 23ai is also not resting. It monitors your database, automatically flagging unusual activity that might indicate performance issues or security threats. This proactive approach means DBAs can resolve potential problems before mushrooming into full-blown catastrophes.

Embracing a thinking future

Oracle Database 23ai isn’t just a database; it’s Oracle’s renewed vision for the future, where AI is seamlessly integrated into the core of data management and databases become pillars of innovation.

The launch of the database release also marks a profound shift in the database landscape. It’s a realization that the database world has changed, and it’s a resounding declaration that Oracle’s ambitions are to lead the charge into this new era. As Tsai confidently states, “It’s gonna get better.”

The new slew of features makes sense for AI-minded customers. Although Tsai notes that the migration path will be smooth and tooled up (it also has a free version), customers will still vote with their wallets. But simplicity is a powerful argument and very persuasive for resource-hungry enterprises.

The pressure is now on competitors to match Oracle's innovation and simplicity. Whatever it is, the race to define the future intelligent database has begun. Game on, DBAs!

Image credit: iStockphoto/AntonioSolano