Kafka Streaming

Review the complete list of supported data entities and data streaming parameters for the Samsara Kafka Connector.

Contents

1. Supported Data Entities
2. Data Streaming Parameters
3. Data Resilience

Supported Data Entities

The Samsara Kafka Connector supports the following data entities in the following categories:

• Asset readings like location (GPS) and speed, and vehicle diagnostics like engine state and fault codes
• Events like AddressUpdated and geofenceEntry

Feedback requested. Samsara plans to support additional data entities over time based on customer feedback. We want to know what additional entities we should prioritize based on your data pipelining objectives and business goals. If you would like to see other data entities supported, please share feedback here.

Data Streaming Parameters

• Data Entities — Select among dozens of data entities detailed below. Each entity can be streamed to a designated Kafka topic in your connected cluster. For each stream subscription, a unique message is produced in the Kafka topic with the data payload for the specific entity. For more information about setting up Samsara’s Kafka Producer see our Product Guide.
• Data Granularity — Data is streamed as it is updated on the Samsara cloud. Exact data granularity depends on the specific entity. For example, when a platform event occurs a message payload is emitted to the designated Kafka topic in real-time. For diagnostic data, please consult this reference for data update frequencies. Note that GPS data updates every 5 seconds when the vehicle is on, and every 5 minutes when the vehicle is off or idle.

Each stream subscription enabled will send messages for one (1) entity to the designated Kafka topic. Entities are not bundled into the same message payload today.

Feedback requested. As you set up stream subscriptions for the available data entities, please share feedback here. We want to hear where you encounter any friction in the setup process or with streaming, or any other feedback you may have.

Data Resilience

If Samsara is ever unable to reach your Kafka cluster (i.e. an outage on your side) our Kafka streaming feature has data resilience built in.

• All time-series entities (e.g. Asset Diagnostics) will recover up to 6 hours of data. If the connected cluster is down, we will retry the connection periodically to backfill up to 6 hours.
• Platform events (see above) cannot be backfilled today; however, if your connected cluster is down we will retry following the same process as for webhooks.

We don’t expect Samsara’s service to go down (learn more about our uptime here) but in the event that happens, no data is lost: the stream subscription will pick up from the last streamed datapoint, until it’s caught back up to current data.

Message Schema Versioning

Message schema version is fixed at the time the stream is created. Like with our API and Webhooks, users of data streaming can assume there will be no breaking changes to subscription payloads.

Security Considerations

Your IT team may need approval from your internal Security team to allow an external producer such as Samsara to write to your Kafka cluster. Learn more about these considerations in our product guide.