Arroyo
AI NativeArroyo is an open-source, Rust-based distributed stream processing engine that runs stateful SQL pipelines (windows, joins, aggregations) with low-latency outputs and can be self-hosted or consumed via Cloudflare Pipelines.
Arroyo is an open-source, Rust-based distributed stream processing engine that runs stateful SQL pipelines (windows, joins, aggregations) with low-latency outputs and can be self-hosted or consumed via Cloudflare Pipelines.
Published May 15, 2026
Searches for G2, TrustRadius, and Gartner Peer Insights did not return a dedicated product page with end-user reviews for Arroyo in this research run. As a result, publicly available user feedback is concentrated in community discussions (Reddit), GitHub issue threads, and individual social posts.
From a Reddit thread discussing the open-source Arroyo engine:
From a LinkedIn post by an individual experimenter:
From Reddit (state backend limitation discussed):
From Reddit (documentation discoverability):
Arroyo’s website provides example SQL pipelines (e.g., Mastodon hashtag analysis, a fraud-like example). These illustrate capabilities—window functions, JSON extraction, Kafka sinks—but are tutorial examples rather than independently verified production case studies and do not include ROI methodology.
Independent articles and technical commentary frame Arroyo as an engine with performance and operational trade-offs; treat these as individual analyses rather than controlled benchmarks.
Get a comprehensive analysis of Arroyo including market position, competitive landscape, adoption trends, and peer benchmarks.
Download full report →Requirement for large, durable in-memory state in the open-source distribution (high-cardinality joins, long-running windows, very large aggregations). The OSS distribution stores state in worker memory, which is an architectural constraint.
Need for comprehensive vendor-attested compliance artifacts (SOC 2, ISO 27001, detailed security whitepapers) provided directly by the product vendor.
Preference for the most conservative, extensively field-tested streaming engine with a very large pool of practitioners; established platforms such as Flink generally reduce platform risk in such cases.
Need for a single-vendor, end-to-end managed Kafka + governance + connectors suite; managed Kafka platforms may be a better single-vendor option.
Primary requirement for a streaming database that maintains continuously updated relational views; a product like Materialize may be more suitable depending on workload semantics.
Cloudflare’s announcement describes Arroyo as the engine behind Cloudflare Pipelines (streaming ingestion and delivery to R2 / Iceberg). This provides evidence of production use within a large infrastructure provider.
Critical view:
Arroyo is an open-source, distributed stream processing engine (written in Rust) that enables teams to build real-time pipelines using SQL. It focuses on stateful stream processing (windows, joins, aggregations) with low-latency outputs and an emphasis on operational simplicity, targeting teams that prefer SQL ergonomics similar to those of data warehouses for streaming workloads.
Arroyo can be self-hosted (single binary / Docker / Kubernetes). The project was acquired by Cloudflare in 2025, and Arroyo is used within Cloudflare Pipelines (managed ingestion and transformation into Cloudflare storage services).
Positioning: marketed as a cloud-native stream processing engine intended to reduce operational complexity compared with JVM-based platforms while supporting stateful workloads and exactly-once semantics.
Target users: