Block storage is for OS and DBs: low latency, attached to instances. Object storage is for files, backups, and static assets: scalable and often cheaper. Use both: block for running systems, object for bulk and archive.
Block storage
- Attached to instances like a disk; OS and apps read/write blocks. Low latency; good for DBs, OS, and apps that need filesystem semantics.
- Sizes and IOPS: Provision capacity and often IOPS; scale by adding or resizing volumes. Backed by SSDs or HDDs depending on tier.
- Use for: Root volumes, DB data, anything that needs random read/write and low latency.
Object storage
- S3-style API: Store objects (files) with keys; no filesystem. Highly scalable; pay for capacity and transfer. Often cheaper per GB for large volumes.
- Use for: Backups, static assets, media, logs, archives. Lifecycle policies for tiering and deletion.
- Not for: OS or DB primary storage; higher latency and different access pattern than block.
When to combine
- Block for running VMs and DBs; object for backups (e.g. DB dumps, volume snapshots exported to object), static files, and cold data. Replicate or copy from block to object for DR.
Summary
Block = disk-like, low latency, for OS and DB. Object = scalable, key-based, for files and archive. Use both: block for live systems, object for backup and bulk.
