Let me ask it another way, is it strictly necessary to use any of these (AWS S3, Azure Blob Storage and Google Cloud Storage) to achieve redundancy?

Good question! lakeFS is not itself a data storage layer. So it uses an underlying storage. It needs one of these object stores. You can of course use many S3-compatible servers instead of S3 itself; we test with S3 as part of our regular testing flow.

Min.io can be installed on K8S and serve as a lakeFS storage layer.

Thank you sou much! So when you plug a PostgreSQL to your LakeFS you are using your Local File System to store objects right?

Here's a quick primer... The [architecture](https://docs.lakefs.io/understand/architecture.html#overview) has a nice drawing of where lakeFS fits in. lakeFS uses an object store for underlying storage and also a database. Big immutable objects go to underlying storage (S3 or S3-compatible, Azure BlobStore, GCS), small mutable objects go to the database (PostgreSQL). It needs both. For non-production use, you can use a regular filesystem as underlying storage; this is called the "local" backing store. It is great for getting up to speed, or for time-limited actions. But it is NOT a supported backend for production.