Although they both offer core IaaS features like virtual machines, storage and databases the leading public cloud providers, Amazon Web Services and Microsoft Azure, take very different approaches in offering cloud services, including at the most basic level of how their data centers are constructed and positioned around the world.
+MORE FROM NETWORK WORLD: What’s behind the Amazon, Microsoft and Google’s aggressive cloud expansions? (With an interactive map!) +
Both companies’ clouds are made up of regions: AWS has 14 and Microsoft has 30. But those numbers aren’t quite an apples-to-apples comparison.
AWS uses what are called Availability Zones (AZs) as the basis for its cloud. Each region in AWS is made up of at least two AZs. Microsoft, instead, just uses regions; it does not guarantee that each region will have multiple data centers. It’s a fundamental difference in the way these IaaS platforms are constructed. And it begs the question: Is one way better than the other?
First of all, it’s important to understand what AZs are why AWS is architected with them. AZs consist of one or more data centers, each with redundant power, networking and connectivity. “As we’ve learned from running the leading cloud infrastructure technology platform since 2006, customers who care about the availability and performance of their applications want to deploy these applications across multiple Availability Zones in the same region for fault tolerance and low latency,” AWS explains on its website. “Availability Zones are connected to each other with fast, private fiber-optic networking, enabling you to easily architect applications that automatically fail-over between Availability Zones without interruption..” Amazon’s credence is to prepare for failure and back up workloads to multiple locations; AZs make that relatively easy.
Regions work just fine for Microsoft
Microsoft takes a different approach. The company does not get into the makeup of its regions. But instead the company has clearly focused on building out a geographically diverse set of regions for Azure. “Geographic expansion is a priority for Azure because it enables our customers to achieve higher performance and it support their requirements and preferences regarding data location,” the company states.
AZs vs. Regions
For most customers, the differences between these approaches will not matter too much, says Jared Wray, former CTO of CenturyLink’s cloud and now an entrepreneur. He says the most common deployment method for enterprise apps migrating to the cloud is to host them in one region and back it up to another (an active-passive setup). In AWS’s case, can be done all in one region across two AZs. In Microsoft or Google’s case another region would be used. Cloud providers offer good enough latency across their regions for most workloads, Wray says.
Where AZ’s shine, he notes, is if you need a low-latency backup, or an active-active backup. AZs within the same region have fast dedicated connections, which are ideal for these apps.
AZs come with some downsides though, Wray adds. If an AZ has an outage, many customers have designed their workloads fail over to another AZ in that same region. If all of the customers in the downed AZ transfer workloads to the healthy AZ, that could theoretically create a crowded AZ, which could impact performance.
To help alleviate this issue, cloud vendors are continually adding new regions. Just this month Amazon launched a new region in Ohio to help offload capacity from it’s Virginia region and satisfy increased demand for their services. Microsoft brought on two new government regions this month too.
At the end of the day, for most customers the AZ vs. region debate will likely not be a deciding factor of which cloud provider to go with. For certain customers, it could be a big factor depending on how latency-sensitive the backup of the app needs to come online. Either way, it’s certainly is something to at least be aware of.