Highlights:

  • SingleStore unveiled improvements to vector search that leverage hierarchical navigable small world, an innovative approach for approximate nearest neighbor search in massive datasets, to boost performance by roughly 40%.
  • Additionally, SingleStore is launching a fully managed private cloud solution that provides the scalability and management capabilities of the database-as-a-service platform.

A real-time database platform developer, SingleStore Inc., integrated Apache Iceberg open-source table format and enhanced scaling and search features. The company’s platform runs analytical and transactional operations in the same engine.

One of the most often used formats for storing various datasets is Apache Iceberg, which is growing in popularity. It offers features like schema evolution, querying historical data, rolling back to previous versions, and transactional consistency across many applications.

SingleStore claimed to handle the frequently complex process of extracting, transforming, and loading data into Iceberg-based data lakes. It claimed that its method enables real-time operation, bidirectional data flow, and low-latency ingestion.

The organization, formerly known as MemSQL, stores data using a distributed, hybrid architecture with disk-based column storage and in-memory row storage. Iceberg integration’s preview is available to the users currently. No date for general availability has been given.

Additionally, SingleStore unveiled improvements to vector search that leverage hierarchical navigable small world, an innovative approach for approximate nearest neighbor search in massive datasets, to boost performance by roughly 40%. According to the company, SingleStore performs between 47 and 100 times quicker for vector operations than the pgvector PostegreSQL extension due to improvements made to the inverted file flat index, a data structure type used in vector search engines for efficient resemblance search.

Moreover, this edition has improved full-text search features, such as fuzzy matching, enhanced phonetic similarity, keyword proximity-based ranking, and relevance scoring. As a result, fewer specialized databases are required to develop real-time applications and generative AI.

New autoscaling features—now in public preview—make performance more predictable by dynamically scaling computational resources. According to the company, consumers can set processor and memory utilization criteria for autoscaling to prevent unexpected charges or consumption.

Additionally, SingleStore is launching a fully managed private cloud solution that provides the scalability and management capabilities of the database-as-a-service platform. To comply with data residency and governance regulations, users can deploy SingleStore in their tenants on the Amazon Web Services Inc. cloud, where the product is now available in private preview.

Since its commencement in 2011, SingleStore has raised more than USD 460 million in the capital. At the end of the previous year, the private corporation reported that its yearly recurring revenue had topped USD 100 million.