Storage and Database Services
Domain 3 - Storage and Database
Questions test which storage class fits a workload and how managed databases handle backup, failover and scaling.
Core rules
- Object Storage is S3-compatible and bucket-based, used for static content, backups and data lakes. Know presigned URLs for temporary access, lifecycle rules for transition/expiry, versioning, static website hosting, and server-side encryption with KMS keys.
- Archive Storage is far cheaper but has retrieval latency - it is for compliance copies, not assets served to users.
- Block Storage is the disk attached to a Server; snapshots are incremental and can create new volumes. NAS provides shared NFS storage for multiple servers.
- Cloud DB for MySQL/PostgreSQL/Oracle/MSSQL is managed: automated backups, a standby for failover in HA configurations, and parameter management through the console instead of SSH.
- Cloud DB for Redis and MongoDB are managed NoSQL options; treat them as VPC-internal endpoints, not internet-facing.
Worked example
A media team stores 40 TB of original footage accessed a few times a year, plus thumbnails served to web users. Reasoning: originals go to Archive Storage (cost first, latency irrelevant); thumbnails go to Object Storage behind CDN. A lifecycle rule that moves objects to Archive after 30 days automates the split, and versioning protects against accidental overwrite.
Traps
- Choosing Archive for content users request directly.
- Assuming a managed DB endpoint is reachable from the internet - it is bound to the VPC subnet.
- Forgetting that deleting a DB instance can also delete backups unless retained.
- Using Object Storage as a POSIX filesystem for applications needing file locks.