One runtime, from ingest to action.

A GPU-accelerated stack carries streaming, warehouse and location data in one format, and scales in data volume, throughput and concurrent users.

Platform overview
Each component shares one data format, so data moves from ingest to action without being copied between layers.
Your sourcesKafkaMQTTDevicesWarehouseROW64Open-source APIsStreamDataWebDashDiagramStream APILive streamsData APIWarehouse dataRAMDBLive stateRow64 ServerRequest-time analyticsLive command centerOperatorsHeadless clientsDownstream systemsYour systemsAlertsTicketsREST calls
Stream API
Stream aggregation and processing from services like Kafka and brokers like MQTT, through the Python Stream API.
A diagram illustrating a cross-filter query system.
RAMDB
An in-memory data store that holds the current state data in one place.
A diagram illustrating the architecture of a RAM database.
Server
A high-speed WebSocket compute server orchestrating bidirectional communication to and from clients.
A collection of icons representing different aspects of computer performance.
Live command center
Browser-based, GPU-accelerated rendering and interactivity through WebAssembly with many different types of data.
A collection of six laptop screens displaying various data and visuals.
APIs
Open-source Python and JavaScript APIs to extend and automate every layer.
Icons representing Python, JavaScript, HTML, and CSS programming languages.
Studio
An intuitive authoring environment usable by operators and engineers alike where dashboards and data models are built.
A computer screen displaying a variety of icons, including a calculator, a map, and a search bar.

Detect, understand and act, without the handoffs.

A separate broker, database and dashboard need glue code and stop at the screen. Row64 goes from event to action with nothing in between.

Detect

Real-time performance everywhere

Data is ingested, analyzed, and rendered in one format, with sub-second latency on continuously changing telemetry.

A blue and white image displaying four different types of computers, each with a unique name. The first computer is labeled as a 'GPU Computer,' the second as a 'Multi-Core Computer,' the third as a 'Multi-Tenancy Computer,' and the fourth as a 'Core Computer.' Each computer is represented by a series of small icons, with the icons arranged in a vis
Understand

Every asset. One live command center.

A GPU-rendered surface combines live telemetry, a vector tiler built for data, warehouse data and unstructured content. Query and understand instantly with a complete picture.

A map of a city with a large body of water, showing the location of various buildings and streets.
Act

From detection to action instantly

Every surface is programmable with Row64's open, API-first architecture.

A computer screen displaying various graphs and data, including a line graph and a pie chart.

No rip-and-replace

Row64 offers flexible deployment patterns tailored to your infrastructure and operational requirements. Common patterns include:

Existing real-time pipelines
The Python Stream API receives data from your existing pipeline and aggregates it into RAMDB.
ROW64Your sourcesIoT sensors · SCADA/OT · WMS · TMSYour RT pipelineKafka · Flink · DatabricksClickHouse · SnowflakeRow64 runtimeShared memory, no copyingLive command centerOperatorsYour systemsSCADA · OMS · CMMSYour data warehouseClean state outEND-TO-END LATENCYSeconds, pipeline dependent
Real-time device direct
Events from devices stream directly into RAMDB. Operators and action systems receive very low latency updates.
ROW64Your sourcesIoT sensors · SCADA/OT · WMS · TMSNo pipeline vendor requiredRow64 runtimeShared memory, no copyingLive command centerOperatorsYour systemsSCADA · OMS · CMMSYour data warehouseClean state outEND-TO-END LATENCYSub-second
Real-time and batch fusion
Simultaneously ingest live telemetry and batch updates from warehouse systems on a defined cadence. Both feeds are written to RAMDB.
ROW64Live streamsIoT sensorsSCADA/OTBatch sourcesSnowflake · DatabricksPostgres · OracleMongoDB · REST APIsimmediateon cadenceRow64 runtimeShared memory, no copyingLive command centerOperatorsYour systemsSCADA · OMS · CMMSYour data warehouseClean state outEND-TO-END LATENCYNear real-time, batch cadence sets the floor

Security that fits your environment

Runs where you run it

Deploy inside your own data center or cloud. Your data stays in your environment.

Federated security

Sign in through your identity provider with OpenID Connect.

Layered access controls

Control who can see which data and views.

SELinux support

Runs on hardened Linux servers with SELinux enforced.

Technology partners

Khronos GroupNVIDIALunarG