Microsoft Azure offers developers access to more data center regions than its competitors, but it was late to the game of offering different availability zones in those regions for high-availability use cases. After a few high-profile issues a couple of years ago, it accelerated its roadmap for building availability zones. Currently, 12 of Microsoft’s regions feature availability zones and, as the company announced at its Ignite conference, both the Canada Central and Australia region will feature availability zones now.
In addition, the company today promised that it would launch availability zones in each country it operates data centers in within the next 24 months.
The idea of an availability zone is to offer users access to data centers that are in the same geographic region but are physically separate, and each feature their own power, networking and connectivity infrastructure. That way, in case one of those data centers goes offline for whatever reason, there is still another one in the same area that can take over.
In its early days, Microsoft Azure took a slightly different approach and focus on regions without availability zones, arguing that geographic expansion was more important than offering zones. Google took a somewhat similar approach, but it now offers three availability zones for virtually all of its regions (and four in Iowa). The general idea here was that developers could always choose multiple regions for high-availability applications, but that still introduces additional latencies, for example.