Regatta announced general availability of its RegattaDB, a distributed SQL database unifying transactional (OLTP), analytical (OLAP), and vector workloads with scale, performance and efficiency.
RegattaDB is a drop-in replacement for Postgres and can scale to tens of thousands of nodes and is intended as a data foundation for AI agent systems. It uses one distributed concurrency model that delivers serializable cross-node consistency and uncompromised performance for all three workloads, obviating the need for for external pipelines or built-in synchronization layers. AI agents, using MCP, access a single source repository for OLTP, OLAP and vector data. They get faster access to consistent data, reducing the risk of hallucinations, missing data and response delays, as well as simplfiying a customer’s data pipeline infrastructure.
CEO Boaz Palgi said: “When we started building Regatta, we had no idea the agent movement would become the next major wave in computing. What we did recognize was that the industry's data architecture had reached its limits. The same fundamental data-layer challenges that we saw years ago still exist today and agents are only magnifying them. We looked at the database as a storage and distributed systems problem and solved it from the bottom up. The result is a fundamentally different database architecture, and it is the only way to truly unify transactional (OLTP), real-time analytics (OLAP), and vector search at the scale, performance, and efficiency modern AI systems require."
RegattaDB’s core concurrency control protocol enables fundamentally different workload types to run natively together across nodes’ boundaries without performance tradeoffs or consistency compromises.
Regatta says RegattaDB is fast, delivering 20 times the performance per footprint of other distributed SQL databases. Using 50 commodity cloud nodes, it executed a distributed JOIN across 20 billion rows while sustaining 50,000 ACID-compliant transactional updates per second on the same dataset in just 3 minutes, with no indexes or data co-location optimizations.
An alternative approach streaming data from operational databases to a data warehouse and then running the query could take hours.
Regatta claims RegattaDB’s architecture enables 400 percent denser workloads per server compared to legacy single node databases, and eliminates an average of 75 percent of hardware, power, cooling, and space costs. An overview of RegattaDB's architecture can be found here and information about how its patented concurrency control protocol works here.
Greg Lavender, former Intel EVP and CTO, and board member at Nutanix and Arista, is joining the Regatta Data Board of Directors to help drive the company's next stage of growth.
Comment
The availability of unified analytic, transactional and vector search databases, multi-model databases, is spreading. For example, Databricks, SingleStore, and SurrealDB. It is thought that the Databricks LTAP (Lake Transactional/Analytical Processing) technology actually involves having two copies of data, one for transactions and a second for analytics.
Palgi posted “it is not a unified data platform. LTAP seems to still remain two engines, but that are sharing a storage layer. Postgres for transactions. Spark for analytics. In other words: the data is in one place. The concurrency model, the execution engine, and the operational surface are not.” Read more about this in a Palgi blog.
Other competitors include Azure Cosmos DB, CockroachDB, Google AllloyDB, TiDB, and YugubyteDB.
There will be implementation differences between them, and they won’t matter so much as the performance and scale of the multi-model database involved. Check out RegattaDB perforrmance here and here.
Bootnote
Regatta has raised $68 million from Lightspeed Venture Partners, 83North, TPY Capital, plus enterprises like Salesforce, Comcast, and Amdocs, and industry angel investors Frank Slootman, Eyal Waldman, Bill Scannell, and Greg Lavender. Regatta’s founding team previously founded or led data infrastructure software companies Topio (acquired by NetApp), XtremIO (acquired by EMC), Storwize (acquired by IBM), and ScaleIO (acquired by EMC, today Dell PowerFlex).