Unveiling SingleStore Pro Max, the Real-Time Data Platform for Apps, Analytics and AI

Say hello to SingleStore Pro Max, the ultimate real-time data platform for applications, analytics and AI. Our latest feature release includes everything from faster hybrid vector search to native data integration services, and the introduction of our new Free Shared Tier.

Metamorphosis

Today is special. Exactly 40 on January 24, 1984, Steve Jobs and Steve Wozniak launched a mass-market computer. Called the Apple Macintosh 128K, it boasted a whopping 128K RAM, a Graphical User Interface and a mouse. The personal computer was born and the world would never be the same again. Apple's latest innovation, the Vision Pro mixed reality headset (128GB version), sold out yesterday — just 18 minutes after pre-orders opened!

What makes this day special for us, at SingleStore? Since its inception over a decade ago, SingleStore has undergone an extraordinary evolution — from a Y-combinator startup with an in-memory database (formerly known as MemSQL) to a general-purpose, cloud-first database hardened for enterprise workloads. The evolution of the product has been accompanied by heady revenue growth — last year, SingleStore surpassed $100M in ARR.

More profound than these many years of growth has been the last 12-18 months. Happily coinciding with the generative AI boom, SingleStore has metamorphosed into a real-time data platform.

What do we mean by "real-time data platform"?

At SingleStore, we believe a real-time data platform is one that is built for all data, all data apps and all data pros.

Real time. A data platform that’s real time is designed for zero ETL across transactional and analytical systems, regardless of the source or destination of data. It is capable of taking in vast amounts of fresh data, making it immediately available for use.
Any data. A data platform handles all kinds of data — from structured data that fits in neat rows and columns, to semi-structured (JSON, BSON, text) and unstructured data like audio, video, PDFs and images. It handles the complete data lifecycle, From creation to integration, visualization and insight.
All applications. The data platform serves all kinds of applications — generative AI, transactional applications, petabyte-scale analytics, front-office analytical applications, etc. Its high concurrency allows it to serve end users, not just back office analytical systems.
All makers and data professionals. A data platform doesn’t just serve back-end and full-stack developers but also ML engineers, data scientists and data analysts. Through the applications built by its customers, a data platform also serves every end-user who interacts with data-driven systems.

In this blog — and at SingleStore’s exclusive event live from San Jose, California — we unveil innovations that are more than 12-18 months in the making: these new feature releases are the amalgamation of more than 200 engineers dedicating years of trial, error and iteration. Size and scale considered, we’ve dubbed this release SingleStore Pro Max. The following graphic illustrates our vision for today, and the future.

At the core of the data platform is SingleStore’s high-performance SQL engine that processes both transactions and analytics (based on our pioneering Universal Storage). New in this core is Hybrid Search, the capability to run both indexed vector and full-text search. Adjacent to the core are cloud services that make SingleStore easier to deploy, manage and scale:
- SingleStore Helios cloud service. The cloud service previously known as Singlestore Helios is now SingleStore Helios — named for the Sun god in Greek mythology. This is the same cloud service existing customers are familiar with– With no changes to existing pricing or contractual terms.
- In-VPC deployment. Coming later this year in preview, the in-VPC deployment offers customers the ability to run SingleStore in their own cloud tenancy
The outer layer of the data platform connects and interacts with other data applications and systems, extraneous to SingleStore.
- SingleStore Kai™ is our API that allows customers to accelerate analytics on their MongoDB® applications with no changes to data or application code
- SingleStore Native Data Integration Services is a suite of ETL capabilities designed to move data in and out of SingleStore
- SingleStore Compute Service allows you to spin up CPU and GPUs to power enterprise AI apps and ML flows on SingleStore

Let’s dive deeper into these exciting new capabilities.

Indexed Vector Search (public preview)

SingleStore now supports vector search using Approximate Nearest Neighbor (ANN) indexing algorithms.

Recall that SingleStore already supports Lucene-based full-text search and exact K-Nearest Neighbor vector search based on dot_product and Euclidean distance measures. Now, with both full-text and indexed vector search capabilities, SingleStore offers fast hybrid search along with the full power of an enterprise SQL database — SQL queries, joins, filters and aggregations + programmability, security, scalability and resiliency. SingleStore’s architecture allows users to plug in relatively quickly to virtually any popular ANN library. We already support the most popular indexing algorithms including IVF, HNSW and their PQ variants:

Inverted File (IVF) is an algorithm for efficient ANN search, partitioning data into coarse quantized clusters for faster retrieval.
Hierarchical Navigable Small World (HNSW) is an efficient, high-recall algorithm for ANN search in high-dimensional spaces.
As an enhancement to both these algorithms, Product Quantization (PQ) can be used. PQ divides high-dimensional space into smaller, quantized subspaces for efficient storage and fast ANN search in large datasets.

With vector search (as with other data processing), SingleStore’s unified architecture enables power with simplicity. SingleStore provides the ability to ingest new vector embeddings and make them immediately searchable. The ability to combine full SQL, vector search and full-text search in one engine eliminates the need for using three special purpose databases. And there's no compromise on performance — SingleStore’s vector search provides excellent throughput and response time.

SingleStore ANN search can be hundreds to thousands of times faster than exact KNN techniques. Keep an eye out for our testing and benchmarks in the near future.

Learn more

Announcing SingleStore Indexed ANN Vector Search
Announcing Vector Data Type Support
Hybrid Search in SingleStore; how it’s better than vector or text search databases

SingleStore Compute Service (private preview)

SingleStore Compute Service brings speciality compute to run Python code and third-party apps next to your SingleStore database. The compute service provides GPUs and CPUs on demand for all kinds of database-adjacent workloads including data flows, vector index building, Python UDFs and AI applications. SingleStore Compute Service is in preview — keep an eye out for further announcements.

Our recently launched SingleStore Job Service is powered by SingleStore Compute Service. This service helps automate data workflows, execute SQL/ Python notebooks or call stored procedures per your defined schedule.SingleStore Job Service + SingleStore Notebooks are the first couple of innovations we've brought to Compute Service. And this is just the start for us: we will be adding UDFs, native data apps and data sharing to compute service.

SingleStore Native Data Integration services

To ease the burden and costs of moving data in and out of SingleStore, we are introducing native capabilities for data ingest and egress for multiple databases and data formats. We currently support CDC-in for MongoDB® (now in GA) and MySQL (public preview), with additional source databases in active consideration.

In a few weeks, we will also release the previw for native ingestion capabilities from Apache Iceberg as well as egress capabilities (CDC out). Our new native data integration suite builds on the industry-leading SingleStore Pipelines. In the near future, Pipelines will also allow you to infer schema as data loads.

With this full suite of data integration services, customers can expect easier, lower-cost migrations while also using SingleStore as a source for other systems including cloud data warehouses and lakehouses.

Learn more

Free Shared Tier

While SingleStore already offers a cloud trial with about $600 in free credits, we are now introducing a new, shared cloud-based free tier that empowers startups and developers everywhere to quickly bring their ideas to life — without committing to a paid plan. The Free Shared Tier allows easy migrations to other low-priced shared tiers (in the future) or to production-fit Standard or Premium Editions.

Learn more

SingleStore Kai™: Now GA

Announced in public preview last year, SingleStore Kai is an API that speeds up analytics for MongoDB® applications by a factor of 100s. It requires virtually no code changes to your application or data transformations from data brought in from MongoDB®. While SingleStore Kai previously supported JSON, it now natively supports BSON and offers industry-leading compatibility with MongoDB® query language.

Coupled with CDC in from MongoDB®, SingleStore Kai is now generally available and ready for both analytical and transactional processing for apps originally built on MongoDB®.

And, there’s more. We’ve implemented the ability for nested array subdocuments that can be pivoted to columnstore, enabling queries on nested arrays to run 20x faster — for both JSON and BSON!

Learn more

More new features developers, analysts and IT Ops teams will love

Database branching. This is a game changer for test-dev and self-service online Point-in-Time Recovery (PITR). Available on SingleStore Helios cloud, database branching creates branches of your database with minimal incremental storage or compute costs. This can be especially useful in test/ dev scenarios when dummy data won’t work, but you don’t want to jeopardize production workloads for testing. It's also a great way to give multiple analysts their own copy of a database to modify as they see fit.
Shallow table copy. Similar to database branching, a shallow table copy is a branch — but rather than that of the entire database, it is a branch of a table alone. It lets you copy huge tables almost instantly, with only a tiny amount of extra storage. It can be used for dev/test, perform table-level backups or enable in-pace data transformation workflows.
Smart DR (cross-region storage DR). In Smart DR, your data is replicated from your primary region to a secondary region asynchronously. Most database recovery solutions require you to run a second hot cluster, meaning you pay double to protect against regional failures. We think there’s a better way — with Smart DR there is no hot cluster, but rather a secondary copy of your data. This is a cost-effective DR solution where data is durable and at the time of the event, you can initiate a failover to bring up your databases and apps in the secondary region in minutes.
Role-Based Access Control (RBAC) for SingleStore Helios Cloud.Now generally available, RBAC ensures database users are assigned permissions based on their role, adhering to the principle of “least privilege.” This enhances security by restricting unauthorized access. It also simplifies the overall process for access control, making it more manageable and scalable as the user base expands.
Query history. You can now extract your query workloads as they occur with the full query text, execution time, query owner and database your queries are being executed on — all in real time, without any impact to the performance expected from SingleStore.
In addition to these product innovations, SingleStore recently announced new integrations and partnerships with Snowflake, ThoughtSpot, Debezium and Presto, making it easier for the world’s builders to bring their ideas to life.

Thank you for reading!

Watch the replay of our announcement event: Meet SingleStore ProMax, the Data Platform.
We invite customers to participate in private previews for these and other new features. Reach out to your account team or drop us a note.
Not yet a customer? Start free with SingleStore today — you can even give our new Free Shared Tier a spin!