EC2 Instance store
- What it is
- Block storage on disks physically attached to the host server running the EC2 instance.
- Included in the instance price — no separate charge.
- The defining property: ephemeral
- Data is lost when the instance stops, hibernates, or terminates, or if the underlying hardware fails.
- Data survives a reboot only.
- There is no snapshot, no backup, no detaching and reattaching elsewhere.
- Why use it anyway
- The highest possible I/O performance — millions of IOPS with no network hop, which Amazon Elastic Block Store (EBS) cannot match.
- Right for data that is temporary or reproducible: buffers, caches, scratch space, temporary processing files, or a replicated node in a cluster that can rebuild itself.
- Not to be confused with
- Amazon Elastic Block Store (EBS) — network-attached, persists independently of the instance, snapshottable, movable within an Availability Zone. The default choice unless you specifically need instance-store speed.
- Where you meet it
- Storage Optimized (D2) and the I/D/H families are built around large instance stores.
- Exam signal
- “Temporary data”, “cache/buffer/scratch”, “highest IOPS” → instance store. Anything that must survive a stop → EBS.