Azure, availability zones, and stability


Generally speaking, a public cloud is more stable than your private servers. After all, in a typical public cloud data center, there are hundreds of thousands of servers. When one server goes down, or even if a few dozen do, it’s no big deal. But, as Murphy’s law dictates: “Anything that can go wrong will go wrong.” So, when outages do occur,  your services can keep running by being shifted to another availability zone (AZ). It can be difficult and it doesn’t work perfectly every time, but it works.

An AZ is a logical data center within a cloud region. A cloud region is a geographical area with multiple data centers and AZs, such as AWS’s US-EAST-1.

Each zone has redundant separate power and private fiber-optic network connections. This way, if one zone fails, your services need not fail with it. A single zone is not a single data center. Typically, a zone is supported by one or more physical data centers. While a single AZ spans multiple data centers, no two zones share a data center.

AWS pioneered AZs and since then most other public cloud services, such as Google Compute Engine and Oracle Cloud, have adopted this approach. For a time, Microsoft took another path: Azure Regions.

In Azure Regions, regions have multiple data centers, but they’re not divided into AZs.  The practical difference is apparent when, as most companies do, apps are deployed in one region and backed up to another. With AZs you do this across two AZs. With Azure Regions you had to do this across regions. Low-latency AZs enable this to be done much more quickly. Most of the time, you wouldn’t notice any difference between these methods, but when something goes wrong–and you need your backup up and running Right Now–it’s a big deal.

Microsoft has recently adopted an AZ model as well. As Tom Keane, Azure’s Head of Global Infrastructure, said, “As part of our commitment to providing customers with a platform for their most demanding, mission-critical workloads, I’m excited to announce expanded capabilities for Microsoft’s global cloud infrastructure.”

Like its public cloud rivals, Azure AZs are fault-isolated locations within an Azure region. They provide redundant power, cooling, and networking. The point of these new AZs is to enable Azure customers to run mission-critical applications with higher availability and fault tolerance to data center failure.

Keane stated, “AZs are now in preview in two regions, East US 2 in Virginia and West Europe in the Netherlands, with plans to offer preview to additional regions in the US, Europe, and Asia before the end of the year including our new France Central region in Paris.”

To make this possible, Microsoft is expanding its data center capacity. Virginia and Amsterdam is just the start. The company plans to add more servers to all of its data centers moving forward.

Availability zones add an extra layer of resiliency in case of a data center-level failure in case of fire or a cooling of flood problem,” Mark Russinovich, Microsoft Azure’s CTO, told Fortune. Microsoft is currently running 36 Azure data centers. It has another eight planned.

For would-be cloud consumers, this will make it easier to compare the big public clouds. For Azure customers, it will provide high-availability for their work loads. And, for Microsoft, it shows–not that there was any question about it–they’re competing as hard as ever with AWS and Google Compute Engine. The overall winner? Cloud users.


Microsoft: The cloud business

Azure Stack will ship soon

Gartner shows us a world of public cloud haves — and have-nots


  1. […] Microsoft regions keep your data safe during an outage, they don’t guarantee quick and easy access. And what’s the point of keeping data safe if you can’t access it when you need it […]

  2. […] Azure, availability zones, and stability […]

  3. […] you use a public cloud, you’ve used availability zones (AZ), but what are they really? How do they fit into regions anyway? Let’s go over the […]

Speak Your Mind


This site uses Akismet to reduce spam. Learn how your comment data is processed.