What’s next for Google Cloud databases? AI inside SQL and more

SQL survey language was the foundation stone of database technology for decades.

But what happens when you bring SQL together with a modern generative AI? This question is Google Cloud Today, the company responds as part of a number of database updates broadcast on Google Cloud.

In the last year, all Google Cloud Database Using AI usage, they added one type of vector support. Google Cloud receives MongoDB compatibility, including the next, multiple databases, including more than one database. Google Bigtable, get supported in the materialized ideas and support support for Oracle Database in Google Cloud.

However, the biggest and most transformation news, at least one database is on the Pattern, AlloyDB database. For the first time Alloydb began In 2022, as an advanced version of the open source postgresql database. At Google in the next event in the summer of 2024, Vector embeddings alloydb, Support for duet AI to activate database migration.

Today is expanded with integration with Alloydb, Google Agenspace, it also debuts in Google Clouds next event. Perhaps it is also interesting, although for the first time within SQL surveys, it is a new AI survey engine that allows SQL statements directly to SQL.

AllOYDB’s API inquiry engine brings a natural language directly to SQL

Google’s new AI survey engine for AlloyDB allows developers to use natural language expressions within the standard SQL inquiries, not only by replacing SQL – SQL.

“We informed the AI ​​survey engine,” Oath GMANS, GM and VP’s VP’s databases on the exclusive interview in VentureBeat. “We will have operators in one SQL request that both of your natural language and basic models and traditional SQL operators can use and bring them together.”

This innovation is an important evolution in the database interfaces. A shortening of structured questionnaires was first submitted in 1973 in 1973. There was a de facto standard for structured database inquiries. The original promise of SQL was to make the database surveys easier to carry out English words in a natural use of the words. Common SQL queries and actions include terms like ‘enter’ and ‘join’, but this is not a very natural language.

“We deliver SQL in a 50-year-old promise to imitate English now,” Gutman said.

The survey engine allows developers to combine accurate SQL syntax with flexible natural language expressions.

Unlike other approaches to SQL translating the natural language, Google’s application connects the language of natural language directly into the question. Google, semantic operators working with the Google Foundation model with traditional related operators in the database engine.

“When SQL came out in 1973, we were amazing, we’re a natural language for survey information and SQL was a kind in this natural language,” said gutmans. “Really, this is now that you are thinking of this, because it is more of SQL, because now SQL can use the natural language as part of your inquiry, but it is still a good structure.”

Agentspace Integration Database Introduction

Google Cloud also combines with the agentspace platform by creating a natural language interface, which is a natural language interface that obtains a database outside of technical specialists in an organization, almost technical specialists in an organization.

Developers and database managers benefit from the AlloyDB AI survey engine, regular work users will use agentspace.

“This is for the average employee in an organization,” said Gutman. “One of the ways to do their work is to have a natural language interface to ask all the enterprise information they really achieve.”

This integration is especially powerful, because the entry is expanding the security. Unlike other natural language database interfaces, Google’s application uses a powerful agentspace platform that knows how much data source, but not about a source of information. This website can be searched, alloydb or other enterprise unstructured data.

Vector search optimization provides measurable work results

Google has dramatically improved alloydb’s vector search opportunities, which optimizes both performance and use. AlloyDB provides 10x fascinated vector search queries compared to the expandable nearest neighbor (Scann) index, hierarchical navigable small world (HNSW) indices.

“AlloyDB saw the vector search commitment in 2024 the nearest neighbor (SCANN) for the AlloyDB index, the nearest neighbor (SCANN) has increased seven times.

This rapid adoption reflects the effects of real business, as it is proven with the experience of the retail giant target. Gutmans noted that the target is using AlloyDB to improve online search experience.

“They use Vector search and use these opportunities to improve accuracy,” he said. “And if you think about 20% of the income, the accuracy of revenues … 20% better targeting means more conversions, more income.”

Real-time handling capability progressed with Bigtable’s materialized views

One of the more technically important ads is a new-resistant new sustainable review features designed for high transmission, real-time applications.

“It’s really a cool ability to really specially,” said Gutman. “Bigtable ClickStream is very used in the meters, there are very low delays and scales as real-time counters for real-time applications.”

Unlike traditional materials that require periodicals require freshness, the implementation of Bigtable is automatically updated.

It eliminates the needs of complex information to facilitate the architecture for real-time analysts, calculating the abilities, and the needs of the complex information to facilitate the architecture.

What does this enterprise mean for the AI ​​adoption

Google’s database accessories offer several urgent advantages to develop AI applications. The AI ​​query engine allows you to learn more easily while maintaining the structure and security of SQL. Optimized vector search provides measurable performance improvements for semantic search applications. Finally, the agentspace integration expands database receipts without demanding SQL expertise.

These innovations for businesses looking at the AI ​​adoption, the infrastructure of the database can only take an active part in the workflows, not maintaining data. The rapprochement of the structure of SQL with the convenience of the natural language creates opportunities for easier applications that use both human and machine intelligence without requiring a complete system redesign.

Perhaps most importantly, Google approaches should not be able to give up existing database investments to accept enterprises’ opportunities for enterprises. When the Gutmans asked SQL whether it was obsolete – “SQL is dead. Long live SQL”.