Our S3 cache is full of holes – We put them there

(bzrk.dev)

5 points | by borlum 5 hours ago ago

1 comments

  • borlum 5 hours ago ago

    At Berserk we use sparse files to keep partial copied of database segment files in our S3 cache. This is a technical walk-through of how and why you might do that.

    Hi I'm Thomas, one of the founders of Berserk. We set out to build a modern telemetry database from the ground up based on object storage. Our goal is to be make very cost-efficient telemetry database with some cool features for working with unstructured and semi-structured logs, metrics and traces.

    AMA