Oracle futureproofed its database with AI vector search and large language (LLM) support in Oracle Database 23c as it melds generative AI applications with enterprise transactional data.
The company, at Oracle Cloud World, also outlined a bevy of next-generation services and products including Oracle Database Appliance X10, Exadata Exascale as well as AI-assisted development in APEX and GoldenGate 23c, which captures, moves and preps enterprise data transactions for private LLM embeddings and analytics.
Constellation Research analyst Holger Mueller said Oracle's database moves set the company up for the future.
""Oracle is leveraging AI at every level of its flagship database product, from operations and assistants, e.g., in the low code APEX platform. But it also is laying the foundation to bridge one of the key challenges of AI today – to bridge the gap of processing between LLMs and transactional data. The solution is vectors fueled by transactional data that can be parsed by the generational AI models, so from a future proofing the new vector DB announcement is a major investment and key step to future proof the relevance of the Oracle Database for the next decade."
- Oracle's MySQL HeatWave additions take aim at AWS with lakehouse support
- Oracle's plan for Cerner: Cloud shift, generative code rewrite
- Oracle, Microsoft launch Oracle services within Azure data centers
- Oracle's Q1 better than expected and Ellison loves generative AI
Oracle argued that Oracle Database has vector database attributes such as scaling, native binary JSON and in-memory column store to vector embeddings. Oracle also introduced a new vector datatype that it will submit to the SQL standards committee, vector operators and vector indexes.
Ultimately, what Oracle is trying to do with vector databases is extend its advantage in operational enterprise data by avoiding data copies, fragmentation and divergence. Multiple algorithms can also be used in the same query. Vector databases have been a hot topic among AI and data vendors.
- C3 AI launches domain-specific generative AI models, targets industries
- MongoDB launches Atlas Vector Search, Atlas Stream Processing to enable AI, LLM workloads
- Databricks Data + AI Summit: LakehouseIQ, Lakehouse AI and everything announced
- Snowflake launches Snowpark Container Services, linchpin to generative AI strategy
Other Oracle Cloud World moves include:
- Using LLMs to generate SQL queries by taking natural language questions and generating SQL queries based on database schema. Oracle said this ability, dubbed NL2SQL, will feature continual pre-training and fine tuning, tight control over data security and governance, integration with Oracle Database Infrastructure and various services including OCI Generative AI Service based on Cohere.
- AI assistants for SQL Developer and APEX.
- More than 300 enhancements for Oracle Database 23c and availability on the Oracle Base Database Service (BaseDB) with automation tools, choice of instance types and dynamic scaling.
- Exadata Exascale, which features an architecture that sits on top of existing Exadata storage systems. Exadata Exascale can pool multiple tenant clusters and databases, isolate security and network for tenants and supports database clones. Exadata Exascale, which requires Oracle Database 23c, is planned for on-premises and cloud deployments.
- Oracle Globally Distributed Autonomous Database for infrastructure, database and data center automation in Oracle Cloud. The company said Oracle Globally Distributed Autonomous Database can be used to address global data distribution and sovereignty requirements, scale horizontally, replicate active/active databases and seamlessly route applications. Informatica announced a bevy of integrations with Oracle's data platform and its Intelligent Data Management Cloud. The data governance integration covers Oracle Autonomous Database as well as Oracle GoldenGate.
- Autonomous Database Elastic Resource Pools. Oracle positioned these Elastic Resource Pools as a family plan for autonomous databases where customers are charged for using the pool not individual databases in pools of 128 ECPU, 256 EPCU and 512 EPCU. Oracle said the approach can lower total costs 4x to 8x and help on-premises customers migrating from over-subscribed database systems.
- Oracle Database Appliance X10, an entry-level engineered system for Oracle Database, with more memory, compute and storage.
- Free developer tools including Oracle Always Free Autonomous Database and Oracle Database Free available as Docker images, GoldenGate 23c Free data integration software with an easier interface use in multiple environments.
Related:
Constellation ShortList™ Artificial Intelligence and Machine Learning Cloud Platforms | Get ready for a parade of domain specific LLMs | Generative AI features starting to launch, next comes potential sticker shock
