Elastic Channels for Snowpipe Streaming reach general availability

Elastic Channels for Snowpipe Streaming provide a server-scaled implicit ingestion path per pipe, enabling concurrent producers to stream directly to Snowflake tables and managed Iceberg tables without managing channel names or offset tokens.

Use cases

Imagine you have concurrent applications sending real-time events and you want to avoid managing channel lifecycles and manual offsets.

• Stream data directly into Snowflake and Iceberg tables.
• Ingest data without manual channel management or lifecycle coordination.
• Use fire-and-forget or waitable append methods in supported SDKs.

How to use

Use version 1.8.0 or later of the Snowpipe Streaming SDK to access the GA API for Elastic Channels in Java, Python, or Node.js.

from snowpipe_streaming import StreamingClient

client = StreamingClient()
client.append_rows(
    table_name="my_table",
    data=[
        {"column1": "value1", "column2": 123},
        {"column1": "value2", "column2": 456}
    ]
)

What happens in this code:

1. Initializes the streaming client.
2. Appends a batch of rows directly to the target table using the managed elastic endpoint.
3. Receives durable per-append acknowledgements.

Limitations and requirements

• Generally available (GA).
• Requires Snowpipe Streaming SDK version 1.8.0 or later.
• Elastic Channels provide at-least-once delivery without strict ordering; use Named Channels for ordered, exactly-once ingestion.

Using it in a pipeline

Removes intermediate staging files and message bus steps, allowing real-time producer applications to feed ingestion rows directly into bronze lakehouse tables with low latency.