Storage

Block storage is a volume you attach to a machine. It exists independently of any instance, survives termination, and grows but never shrinks.

There is no object storage on this platform — no S3-compatible endpoint and nothing that pretends to be one.

Block storage resources

User guide What a volume is, root disk against attached volume, sizing for spinning disks, what you must do yourself, and how durable this really is.

Developer guide Driving storage from code: the state machine, the retry that bills you twice, parameters that are refused rather than ignored, growing a volume end to end, and the guest-side half no API can do for you.

Volume types The one type we sell, the performance it delivers, how that figure was measured and divided, and where it is not yet enforced. Read this before you put a database on it.

API reference Every volume operation in detail — create, describe, attach, detach, modify, delete and status — with parameters, responses, errors and worked examples.

Snapshot actions Point-in-time copies, how to restore from one, what a snapshot really guarantees — and the pool capacity it quietly consumes, since nothing expires them.

Storage developer guide The block-storage service’s own API, for tooling that speaks it and for the things the EC2 model has no way to express.

Before you depend on it

One box, one pool, one class of disk — 7200 rpm SATA, quoted at 48 IOPS and 11 MB/s per volume. The pool is not mirrored, there is no cross-host replication, and there are no provider backups. Snapshots live in the same pool as the volumes they protect.

Treat a volume as durable against a process crash, not against a hardware failure. Copy anything you cannot lose off the platform.

Related

Compute — the machines volumes attach to.

Quotas and limits — including the storage quota that is not set.