Actually, my question not about possibility, but is there any value in it? May be any message queue or k8s cluster with StatefulSets is easier to operate? What do you think?

It is an interesting idea. What troubles me is the read/write rate in a real production environment. If it is bound by S3, a buffer will eventually be needed and then it all become very complicated. At least it is how it looks on the surface

You are right, the time of a single write/read is Achilles' heel. But if we are talking about parallel processing it can keeps several hundreds milliseconds delay and provides enormous throughput.

Why it matters:

→ 3x faster than nearest competitors. Rust + Apache DataFusion + Arrow do the heavy lifting.

→ Exactly-once delivery with WAL. No data loss, no duplicates - guaranteed.

→ Your data stays in open formats you control. No vendor lock-in. Ever.

→ Query logs through Loki-compatible API - works with your existing Grafana setup.

→ Ingest via standard OpenTelemetry (gRPC/HTTP). No proprietary agents.

→ Your data is in Iceberg tables - query it with Trino, Spark, Flink, DuckDB, or any tool in the Iceberg ecosystem.

→ Your observability data becomes a first-class asset of your data lake alongside business analytics.

→ Scale ingest, query, and storage independently. Full compute-storage separation provides stateless deployment.

> 3x faster than nearest competitors

Wow, that's awesome! Do you have any public benchmarks?

There will be a series of articles on Linkedin https://www.linkedin.com/company/icegatetech in the near future, where we will tell you more about the technical side and show benchmarks.