Storage Tiering
Storage tiering is the practice of placing data on different storage tiers based on how often it is used. Hot, frequently accessed data goes on fast storage. Cold, rarely accessed data moves to cheaper, slower storage.
How tiering saves money
Not all data is equal. The last hour of metrics gets queried constantly. Data from eight months ago gets touched once in a blue moon, usually for an audit or a long-range trend. Paying premium storage prices for both is wasteful.
Tiering solves this by matching storage cost to access pattern. Recent data sits on fast storage for low-latency queries. Older data ages out to cheap object storage, where it stays queryable but costs a fraction as much. Done well, this cuts storage bills sharply without giving up access to history.
The key is that cold data should still be queryable, just slower and cheaper, not deleted or locked away.
How Arc handles Storage Tiering
Arc keeps data as Parquet on object storage by default, which is already the cheap tier, and supports tiering so hot and cold data land in the right place automatically. You keep long history queryable without paying hot-storage prices for all of it.
Arc is a high-performance columnar database. Open Parquet on storage you own, single Go binary, production-ready in 30 seconds.