A payment API must keep serving requests if one datacentre in its Azure region loses power. Legal advice is that the data may not leave the region. Which arrangement satisfies both constraints?
Microsoft Certified: Azure Fundamentals (AZ-900), objective architecture-and-services. Azure architecture and services medium
Machine-checked — no person has signed for it. This question was read against the source cited below by an automated pass, which found no contradiction. That is a weaker claim than it sounds: the same kind of process wrote the question, so it can confirm its own mistake.
Treat it as a good draft rather than as settled fact, and read the source below before you rely on it. It is not used in mock exams here — only questions a person has signed for are.
The options
Correct Deploy the API across availability zones within that region
Correct. Availability zones are physically separate datacentres inside one region, each with its own power, cooling and networking, so losing one leaves the others serving. Nothing crosses the region boundary.
Not correct Deploy a second copy of the API into the region's pair
Wrong. A paired region does give you independence from a datacentre failure, but it is a different region in a different part of the geography — which is exactly the data movement the legal advice rules out.
Not correct Put the virtual machines into an availability set
Wrong. An availability set spreads instances across fault and update domains inside a single datacentre, so it survives a failed rack or a host reboot but not the loss of the whole datacentre.
Not correct Move the workload to a larger virtual machine size
Wrong. A bigger machine is still one machine in one datacentre. Capacity is not availability, and scaling up changes neither the failure domain nor the number of them you occupy.
Why
Region, availability zone and datacentre form a hierarchy, and each level isolates a different failure. A region is a set of datacentres inside a latency-defined perimeter; an availability zone is one or more of those datacentres with independent power, cooling and networking; an availability set only separates instances inside a single datacentre. Match the failure you fear to the level that isolates it: rack or host to an availability set, whole datacentre to zones, whole region to a second region.
Where this comes from
- Cited
- Microsoft AZ-900 study guide skill area architecture-and-services.core-architectural-components
Practise this
Reading one question is not practice. The trainer will draw a set from objective architecture-and-services and space the ones you get wrong.
Practise Microsoft Certified: Azure Fundamentals (AZ-900)
More questions on this objective
- An image-resizing routine runs for about two seconds whenever a file lands in a storage container — perhaps two hundred times on a busy day, and not at all on a quiet one. The team wants to pay for the work done and maintain no servers. Which compute option fits? machine-checked
- A fifteen-year-old accounting application needs a kernel-level driver, a scheduled task that edits the registry, and a runtime version nobody supports any more. The business wants it running in Azure this quarter with as few code changes as possible. Where does it go? machine-checked
- A team runs nine small services, each with its own conflicting library versions. They want each one packaged with its dependencies, starting in seconds, several to a host, without a separate operating system per service. Which compute type are they describing? machine-checked
- Which TWO of these statements about Azure compute types are true? machine-checked
- A public website runs on four identical virtual machines behind a load balancer. Traffic triples while a television advert airs and falls back an hour later. The team wants instances added and removed automatically against CPU, all built from one image, with nobody clicking. Which option is designed for that? machine-checked
- Three virtual machines form a quorum-based cluster in a region that has no availability zones. You want them placed so that neither a single rack losing power nor a single batch of planned host maintenance can take all three at once. What do you configure? machine-checked