Storage and Data Management
What to know
Know the storage families: object (unstructured, HTTP-accessible, eventually or strongly consistent depending on provider), block (attached to a single instance, low latency), and file (shared across instances). Understand durability vs availability, storage tiers (hot/cold/archive) and lifecycle policies for cost. Databases split into relational (ACID, joins, schema) and NoSQL (key-value, document, wide-column) with trade-offs around consistency and scale. Cover backup strategies, snapshots, replication (synchronous vs asynchronous), and data migration tools. Encryption at rest and key rotation belong here too.
Traps
Durability (probability data survives) is not availability (probability the service responds); questions swap these. A frequent trap is choosing block storage for shared multi-instance access - file or object storage is correct. Another is picking a relational database for massive horizontal key-value scale where NoSQL fits, or vice versa when transactions are required. Watch for lifecycle rules that delete data too early versus transition to a cheaper tier. Also remember snapshots are incremental and stored separately from the volume.
Worked example
A media company stores years of logs rarely accessed but must retain them for seven years. The answer is object storage with a lifecycle policy transitioning to an archive tier after 30 days and expiring at seven years - not keeping them on block volumes. Rule of thumb: choose storage by access pattern and sharing need, then apply lifecycle rules to control cost.