Skip to main content
If your infrastructure already moves events through Kafka, RabbitMQ, or SQS, you don’t need to build a new ingestion path for onchain data. Data Feeds supports familiar streaming interfaces, so you can consume feeds with the client libraries your team already runs:
  • Kafka-compatible: consume feeds using Kafka client libraries.
  • RabbitMQ / AMQP: consume feeds using standard AMQP clients.
  • SQS-compatible: poll feeds through an SQS-style HTTP API.
Which interfaces are available can depend on the feed and current availability, confirm the protocols you plan to use when you onboard.

​
What you receive

Each message carries block-level decoded data: the block’s token transfers, swaps, price updates, balances, NFT activity, approvals, and the raw block/transaction/log data, laid out as typed arrays. You consume decoded records, not raw RPC payloads. See The block array for the exact layout.

​
Reorg handling

Streaming consumers see blocks as they land. Each block records its position and parent, so when the chain reorganizes, your consumer (or a downstream process) can detect the rollback and correct affected records. This differs from the database sinks, where recipes apply reorg corrections for you. See Reorgs for how rollbacks are signaled.

​
Streaming vs. the sink

You can also combine them: land ready-to-query tables with the sink while streaming the same feed into an existing pipeline. It’s the same data lake underneath, see the access methods overview.

​
Get connected

Connection details, authentication, and per-feed protocol availability are provided during early-access onboarding.