SingleStore review

SingleStore is a database engine offered as a managed service. It combines online transaction processing and analytics in one HTAP system, so applications can query data as it is written rather than relying on a separate warehouse copy. It sits in the data layer beneath applications, dashboards and AI agents that need current business data.

The engine supports relational SQL, JSON, time-series, full-text, vector and geospatial data. The homepage describes millisecond response as its operating goal and says data can be written and queried in the same second. Those are product claims, not a guarantee for every schema or workload; validate latency and throughput with your own query mix.

For AI use, Aura Intelligence includes Analyst, which takes plain-English questions and executes planned queries against the live database, and a Context Engine that maps schemas, relationships and definitions. Iceberg connectivity links operational data to Apache Iceberg formats.

The integration surface named on the homepage is SQL: SingleStore describes a relational database, and its Shared plan includes SQL and programmability features. Analyst adds a natural-language path for querying the live database. The homepage does not name a REST API, CLI, language SDK or specific driver, so confirm the supported connection methods and SDKs in the vendor’s developer docs before choosing a stack.

The product also connects to Apache Iceberg, which gives teams an open-format path for data exchange. The homepage does not enumerate other integrations or provide implementation details for that connection. Ask the vendor which Iceberg catalogs, formats and workflows are supported for your deployment.

For evaluation, confirm that the documentation covers your language, connection pooling, schema changes, retries and observability needs. Teams moving an existing application should test its SQL features and compatibility rather than assume a drop-in replacement for another database. Database branching is listed on Standard, while Shared includes monitoring and self-support.

SingleStore’s Shared plan costs $0 per month and includes one free workspace, a shared workspace, SQL and programmability features, monitoring and self-support. The vendor does not publish a free-tier usage limit in the pricing details here, so check workspace capacity, compute and storage limits before using it for a sustained workload.

Managed production editions bill by the hour. Standard starts at $0.99/hr and lists read replicas, database branching, 30-day monitoring and a 99.9% or 99.99% SLA. Enterprise starts at $1.49/hr and lists online PITR, Smart DR, audit logging and CMEK. These are starting rates, not fixed monthly totals: compute hours and storage usage affect the bill, and the page lists no annual rates.

Optional SingleStore Flow has a pricing example of $3.96 per credit and $0.02/GB processed. Storage examples list $0.023/GB and $0.025/GB, varying by example and region. SingleStore says it does not charge for data ingress, though cloud-provider charges may apply.

Reddit commenters say SingleStore’s core engine is open source and can be run in a container, but the licensing terms are not established in those discussions. Confirm the current licence and permitted production use with the vendor before treating self-hosting as an option. The available pricing details describe managed Shared, Standard and Enterprise editions; they do not establish a supported self-hosted production offering.

A SingleStore employee says the storage format is not open, which can make migration harder than moving data between systems that share an open storage format. The homepage presents Apache Iceberg connectivity as a way to connect operational data to an open format, but that does not establish a complete export or recovery path. Ask how to export data, schema and change history, and test that process before storing critical workloads. The homepage’s claim of avoiding catalog or vendor lock-in should be assessed against those practical migration requirements.

The pricing page lists a 99.9% or 99.99% SLA with Standard, depending on the selected offering, and Enterprise includes online point-in-time recovery, Smart DR, audit logging and customer-managed encryption keys. Confirm which SLA applies to your configuration, how recovery objectives work, and whether the security controls meet your requirements. The listed 30-day monitoring is a Standard feature; Shared includes monitoring but no stated retention period.

The homepage positions SingleStore for millisecond responses, but real performance depends on the workload. One Reddit user reports comparable performance to PostgreSQL in a test but says SingleStore had fewer features and less compatibility. Another commenter cautions that a benchmark using simple, highly compressible data may not predict other workloads. Build a proof of concept with representative data, write rates and queries.

Check transaction semantics, capacity scaling, storage costs and export procedures before production. A Reddit commenter affiliated with SingleStore says regular transactions provide READ COMMITTED isolation and SNAPSHOT is available for backup; verify current behavior in the vendor’s docs.