Blog#

Icechunk Adopted by the National Weather Service: Earthmover Joins Booz Allen on NWS CIRRUS

June 4, 2026 — Icechunk will serve as a core data format for NWS CIRRUS, NOAA's new cloud-based operational data platform for the National Weather Service.

Old format, no problem! Cloud-optimizing the GOES-16 archive as Virtual Zarr

June 2, 2026 — Using VirtualiZarr and Icechunk to expose a massive archive of GOES-16 satellite imagery as a single cloud-optimized Zarr store, without copying any data.

Ship fast, break nothing: Engineering rigor in Icechunk with cross-version stateful compatibility testing

May 21, 2026 — Using third-wheel and Hypothesis to interleave Icechunk 1.x and 2.x in a single stateful test, and the bugs that fell out.

Variable length chunks in Zarr

May 5, 2026 — A new Zarr extension, the rectilinear chunk grid, lets you align chunk boundaries with the natural structure of your data instead of forcing a regular grid.

Ship fast, break nothing: Engineering rigor in Icechunk with concurrency & fault injection testing

April 28, 2026 — Concurrency permutation testing with shuttle and fault injection with toxiproxy to harden Icechunk 2.

Designing a data pipeline for our highest-resolution dataset yet

April 22, 2026 — CarbonPlan on using Icechunk's transactional writes to build a fault-tolerant, idempotent pipeline for their highest-resolution wildfire risk dataset.

Ship fast, break nothing: Engineering rigor in Icechunk with property and stateful testing

April 22, 2026 — How property-based and stateful testing keep Icechunk reliable as it evolves.

Announcing Icechunk 2: Better Consistency, Performance, and Reliability for Tensor Storage

April 9, 2026 — Release announcement for Icechunk 2, covering node rename, chunk reindexing, and performance wins.

Cursed venvs, Confident Releases: Testing Icechunk Across Major Versions

March 20, 2026 — Introducing third-wheel, the tool used to test cross-version compatibility for the Icechunk 2 release.

Evolving our Tensor Storage Engine: A Preview of Icechunk 2

December 2, 2025 — Preview of Icechunk 2 features and the migration path from 1.x.

Zarr Everywhere

October 13, 2025 — Development Seed's overview of their work on cloud-native array tooling, including Icechunk, VirtualiZarr, and Zarr v3 adoption across the ecosystem.

From 10 Minutes to 10 Seconds: How Woods Hole Scientists used Icechunk to Optimize Ocean Data Access

October 8, 2025 — Case study on converting OPeNDAP-served NetCDF to Icechunk for dramatic latency improvements.

Multi-Player Mode: Why Teams That Use Zarr Need Icechunk

August 25, 2025 — Why atomic updates, consistent snapshots, and Git-like version control matter for teams writing Zarr collaboratively.

Icechunk 1.0: Production-Grade Cloud-Native Array Storage Is Here

July 10, 2025 — The stable 1.0 release: transactional safety, versioning, and virtual file references.

Everything you need to know about Icechunk garbage collection

May 30, 2025 — Practical guide to garbage collection and expiration for reclaiming storage.

Icechunk: Efficient storage of versioned array data

May 14, 2025 — How Icechunk stores versioned data efficiently by avoiding chunk duplication.

Learning about Icechunk consistency with a clichéd but instructive example

April 23, 2025 — Transaction semantics and conflict detection, illustrated with a bank-transfer analogy.

Exploring Icechunk scalability: untangling S3's prefix story

April 10, 2025 — Demonstrating that Icechunk scales to hundreds of thousands of requests per second on S3.

Solving NASA's Cloud Data Dilemma: How Icechunk Revolutionizes Earth Data Access

March 27, 2025 — Case study on the Earthmover and Development Seed NASA pilot, enabling 100x faster cloud-native access to archival Earth science datasets without costly data migration.

Announcing Icechunk!

October 15, 2024 — The original announcement introducing Icechunk as an open-source transactional storage engine for Zarr.