Azure Storage on AZ-900 is not an architecture puzzle — it is a set of vocabulary decisions the exam checks over and over. Learn to make each one quickly and you have banked most of the storage marks. This summary is the shortcut: the facts that get tested, in the order you decide them.
First decision: which service?
The exam almost always starts by describing a workload and asking which storage service fits. The right-hand panel of the diagram above is the whole answer.
- Blob Storage — object storage for unstructured data: images, video, backups, logs, anything you count in the millions. This is also the only one of the three that has access tiers.
- Azure Files — a managed file share several machines mount over SMB or NFS. The tell is “a shared drive” or “multiple machines need the same files.”
- Managed disks — block storage attached to a single VM as its OS and data drives, persisting across stop and start. The tell is “a virtual machine needs a durable drive.”
One line to carry in: objects → Blob, shared file share → Files, a VM’s drive → Disk.
Second decision: which Blob access tier?
Once you are in Blob Storage, the exam tests the cost-versus-access trade shown on the left of the diagram. Read the ladder top to bottom:
| Tier | Best for | Read speed |
|---|
| Hot | Data accessed frequently | Instant |
| Cool | Infrequent, kept ≥30 days | Instant |
| Cold | Rarely accessed, kept longer | Instant |
| Archive | Long-term, almost never read | Offline — rehydrate first |
Going down the ladder, storage gets cheaper but access gets slower and pricier. The one fact that trips people: Archive is offline. You cannot read an archived blob directly — you rehydrate it back to Hot or Cool first, and that takes time. Pick the tier by how often you read, not by the sticker price of storage.
Third decision: how redundant?
Redundancy is a separate axis from tiers — how many copies exist and how far apart:
- LRS (Locally Redundant Storage) — three copies in one datacenter. Cheapest, but a whole-datacenter loss takes the data.
- ZRS (Zone-Redundant Storage) — copies across availability zones in the region. Survives one datacenter failing.
- GRS (Geo-Redundant Storage) — copies in a second, distant region. Survives a regional outage.
The pattern is simply reach: same building, across zones, across regions. Match the option to the failure the question describes — a zone failure wants ZRS, a regional disaster wants GRS. Reaching further than the scenario needs is the classic distractor.
Storage accounts, briefly
Blobs, file shares and other data live inside a storage account — the top-level container that holds a name, a chosen redundancy level and access settings. You do not need its internals for AZ-900, only that it is the wrapper you create first.
How to use the questions below
Answer each one before you open the explanation, and when you miss, read past the correct option to why the others are wrong — the distractors here are the right service or tier offered for the wrong reason. Getting the reason right is what survives into the exam. For the wider service map, see the Azure architecture and services domain and the compute summary; for one-page recall, the AZ-900 cheat sheet.
Updated for AZ-900 (July 2026). The sample questions above are our own work
in the style of the exam — not real exam items. The exam itself is set and marked
by the certification body.