Cloudflare Worldedge community field guide

CURRENT EDITION / 2026-10

Basin and K2: from event stream to an open lakehouse on R2

Basin is generally available as Apache Iceberg tables on R2; K2 adds a durable ordered log in public beta.

ISSUE
34 / 34
FIRST PUBLISHED
SOURCE CHECKED
Archive integrity

The edition month is a curriculum position. This edition was written from first-party sources published in 2026-10 and was itself first published on 2026-10-21; editions after the retrospective run carry their own publication date.

01 / DECISION

When this primitive earns a place

Keep Analytics Engine for metrics and Queues for task fan-out; add K2 when several independent consumers must replay one ordered log, and Basin when questions need SQL over large, open-format history.

02 / ARCHITECTURE

Build the smallest defensible path

Run the Basin Pipelines setup to create a stream, an SQL transformation, and an Iceberg table on R2; let Basin Catalog handle compaction and snapshot expiration; query the warehouse with Basin SQL. On K2, producers append records and subscriptions deliver them to one consumer group or to several independent consumers.

  1. 01Name the contract

    Keep Analytics Engine for metrics and Queues for task fan-out; add K2 when several independent consumers must replay one ordered log, and Basin when questions need SQL over large, open-format history.

  2. 02Add one primitive

    Run the Basin Pipelines setup to create a stream, an SQL transformation, and an Iceberg table on R2; let Basin Catalog handle compaction and snapshot expiration; query the warehouse with Basin SQL. On K2, producers append records and subscriptions deliver them to one consumer group or to several independent consumers.

  3. 03Capture failure evidence

    Post a sample event to the stream ingest endpoint, then read the same rows back with a Basin SQL query against the warehouse; on K2, confirm that each subscription receives the records it should and that a replay returns them in the same order.

03 / REPRODUCE

Evidence before confidence

Post a sample event to the stream ingest endpoint, then read the same rows back with a Basin SQL query against the warehouse; on K2, confirm that each subscription receives the records it should and that a replay returns them in the same order.

npx wrangler basin pipelines setup

04 / BOUNDARY

The production boundary

Basin is generally available, but K2 is public beta on the Workers Paid plan, capped at 10 GB per account with retention up to 30 days. Basin SQL is analytical, not transactional, so it does not replace D1; each Basin Pipelines stream ingests up to 1 GB/s. Cloudflare Pipelines, R2 Data Catalog, and R2 SQL are now Basin Pipelines, Basin Catalog, and Basin SQL, and resources created under the old names keep working.

05 / FIRST-PARTY SOURCES

Keep first-party sources authoritative

Product behavior, limits, pricing, and availability can change. Re-check these sources before acting.

01 / SOURCECloudflare Basinhttps://developers.cloudflare.com/basin/02 / SOURCEBasin is now generally availablehttps://developers.cloudflare.com/changelog/post/2026-10-01-basin-ga/03 / SOURCEBasin Pipelines getting startedhttps://developers.cloudflare.com/basin-pipelines/getting-started/04 / SOURCECloudflare K2https://developers.cloudflare.com/k2/05 / SOURCEBasin Pipelines stream ingest limithttps://developers.cloudflare.com/changelog/post/2026-10-01-stream-ingest-limit-increase/