Cloudflare Rebrands Its Data Platform as Basin and Takes Iceberg Pipelines, Catalog and SQL to General Availability
Cloudflare renamed R2 SQL, R2 Data Catalog and Pipelines as Basin SQL, Catalog and Pipelines and moved them to GA; R2 egress stays free, but pipeline sinks and queries are metered.
Overview
Cloudflare Data Platform received a new brand name, Basin, and was promoted from beta to general availability on Thursday, October 1, according to The Register. Under the new name, Cloudflare Pipelines, R2 Data Catalog and R2 SQL become Basin Pipelines, Basin Catalog and Basin SQL respectively. The Register notes that the occasion was Cloudflare Birthday Week.
The three products together turn an R2 object-storage bucket into an Apache Iceberg lakehouse: streaming data goes in through Pipelines, table metadata is tracked by the Catalog, and queries run through Basin SQL or an external engine.
What Each Piece Does
- Basin Pipelines. Per Cloudflare’s documentation, the service is meant to “Ingest, transform, and load streaming data into Apache Iceberg or Parquet in R2,” and it is available on the Workers Paid plan. The docs say it “ingests events, transforms them with SQL, and delivers them to R2 as Iceberg tables or as Parquet and JSON files.”
- Basin Catalog. The Catalog docs describe it as a managed Apache Iceberg data catalog built directly into an R2 bucket that exposes a standard Iceberg REST catalog interface. The same page states that existing configurations and APIs continue to work and will be deprecated in the future.
- Basin SQL. According to the Basin SQL docs, it is Cloudflare’s serverless, distributed analytics query engine for Iceberg tables stored in Basin Catalog. The page says existing R2 SQL APIs continue to work and will be deprecated in the future, and shows queries being run with a
wrangler basin sql querycommand.
Cloudflare CTO Dane Knecht said, as quoted by The Register, that “we are bringing the same serverless model that developers expect from Cloudflare to analytics: no clusters to manage, no unnecessary data movement, and open standards that keep customers in control of their data.”
Open Standards and Outside Engines
The Register describes Basin’s appeal as resting on a lack of lock-in, because it relies on Apache Iceberg, and on a lack of egress fees. Data stored through R2 Data Catalog can be processed with Basin SQL or with external query engines such as DuckDB, PyIceberg, Snowflake, Spark, StarRocks or Trino, according to the same report. The Catalog docs add that Iceberg is already widely supported by engines including Apache Spark, Trino, Snowflake, DuckDB and ClickHouse.
Where the Fees Are
“Zero egress” applies to R2 itself: R2’s pricing page says there are no charges for egress bandwidth for any storage class. The Register’s headline framing, “fewer fees,” is qualified in its own report, which says that Cloudflare charges for delivering data to Pipelines sinks after the included allowance when streaming data is loaded into Iceberg or Parquet in R2.
The documentation lays out three separate meters:
- Pipelines. The Pipelines pricing page says charges fall on two dimensions, SQL transforms and sinks, and that ingress into a pipeline stream is free. On the Workers Paid plan, 50 GB per month of SQL transforms and 50 GB per month of sink output are included. Additional SQL transforms cost $0.04 per GB. Sink output to R2 costs $0.03 per GB in JSON format and $0.06 per GB in Parquet or Iceberg format, measured on uncompressed data. The page also says future stateful operations such as aggregations, joins and windows may be priced separately.
- Catalog. The Catalog pricing page lists 1 million catalog operations per month as included, then $9.00 per million. Compaction, which only incurs charges when turned on for a table, includes 10 GB processed per month and then costs $0.005 per GB, plus $2.00 per million objects after the first million. The page says snapshot expiration is free of charge.
- SQL. The Basin SQL pricing page says the single billing dimension is compressed data scanned, with 10 GB per month included and then $0.0025 per GB ($2.50 per TB). It sets a 10 MB minimum per query. The page says queries that fail from a system or syntax error, or that fail mid-execution from a runtime error, are not charged, and that it requires an active R2 subscription.
Standard R2 storage and operations charges apply on top. R2 standard storage is listed at $0.015 per GB-month on the R2 pricing page.
The Pipelines pricing page includes a worked example: a pipeline ingesting 500 GB of event data per month, with a SQL transform that reduces output to 300 GB written to a Basin Catalog Iceberg table, totals $33.00 for the pipeline itself ($18.00 for 450 billable GB of SQL transforms and $15.00 for 250 billable GB of Iceberg sink output), per the pricing documentation.
What We Don’t Know
- The sources reviewed do not say how many customers are running the formerly beta products in production, or give performance benchmarks for Basin SQL.
- The documentation does not say when the legacy R2 SQL and R2 Data Catalog APIs will be deprecated, only that they will be “in the future.”
- The Register’s report quotes customer endorsements from Cloudflare’s announcement; those are vendor-selected and were not independently verified.
Analysis
The launch is largely a rename plus a status change, with the existing APIs kept working. The practical question for adopters is the cost model: egress from R2 is free, but the pipeline, catalog and query layers each carry their own usage-based charges, which is the point The Register’s report makes with its “And yet there are fees” framing. Teams comparing Basin against hyperscaler lakehouse stacks will need to model sink volume, compaction and bytes scanned alongside storage.