Storage and Data Management
What to know
Alibaba Cloud storage splits into object (OSS), block (cloud disks/ESSD attached to ECS), and file (NAS). OSS uses buckets and objects with storage classes (standard, infrequent access, archive) and lifecycle rules to transition or expire data. Cloud disks come in performance tiers and support snapshots for backup; some support multi-attach. NAS offers shared file access across instances. For databases, know RDS (managed relational) with read replicas, backups, and high-availability editions. Understand durability vs. availability, and that snapshots are incremental and stored separately from the source disk.
Traps
Questions test storage class selection: archive is cheapest but has retrieval latency and minimum-duration charges. Lifecycle policies automate transitions but are not instant. Confusing block vs. object storage is common - block disks attach to one instance (or a limited multi-attach), while OSS is accessed via API/HTTP. Another trap: assuming a snapshot alone is a full disaster-recovery plan, or that read replicas accept writes. Watch for minimum storage duration penalties when deleting archived objects early.
Rule of thumb
Pick storage by access pattern: frequently accessed unstructured data → OSS standard; shared files across many instances → NAS; low-latency disk for an OS or database → block disk. Use lifecycle rules to move cold data to cheaper classes.
Worked example
Logs are hot for 30 days, rarely read for a year, then deletable. Store in OSS standard, add a lifecycle rule to transition to infrequent access after 30 days, then expire after 365 days.