In this post, we’ll explore and compare the AZs offered by the top 3 major cloud service providers, with the goal to help you choose the best option in your cloud migration planning journey.
Comparing leading cloud providers has certainly been a recurring practice on our blog. Check out this other post of ours to help you differentiate between pricing structures at all three CPSs.
That being put there, let’s dive into the topic that we’re dedicating this article to: global infrastructures.
The global cloud infrastructure is a network of distributed data centers spread across various international locations.
The decision of a cloud service provider to set up a data center in a region is based on several factors. CSPs often factor in per-capita demand and income ratio before setting up a Region. Hence, population concentrations in the developed world are more likely to catch the attention of CSPs versus those in the developing world.
Therefore, as far as major cloud vendors go, there are far more Regions in North America and Western Europe than in South America and Africa. GCPs tend to divide these regions into the so-called Availability Zones (AZs) or simply Zones (if we go by Google Cloud nomenclatures).
For the sake of clarity, let’s dive into some brief definitions:
Regions: A Region is a physical location you can host your cloud resources in. Regions are autonomous and resource-exclusive, meaning resources available in a specific Region are inaccessible from other Regions.
Inter-Region communications happen over IP. A Region contains one or more AZs.
CSPs tend to name Regions after the geographical locations they are in. For example, AWS maintains US West Region in California and US East Region in Ohio.
Availability Zones (AZs): An AZ is one of the locations in a Region. A Region may contain more than one Availability Zone. Most cloud vendors offer at least 3 AZs per Region.
Each AZ points to one or multiple physical data centers in the Region. For safety reasons, one AZ is isolated from other AZs and they communicate via low-latency fiber-optics links.
Users consider putting resources in various Availability Zones to ensure isolation and reduce risks.
It’s interesting to note that some types of resources stay in an AZ while others stay in a Region.
What Cloud Service Providers have to offer in terms of global infrastructure?
Global cloud infrastructure grew by 32% to reach a staggering $39.9 billion in the 4th quarter of 2020, following peaked customer spending with the major cloud vendors: Microsoft Azure, Amazon Web Services (AWS), and Google Cloud Platform (GCP). Total spending was $3 billion higher than Q3 2020 and roughly $10 billion higher than the same quarter last year, according to Canalys data.
This spike certainly was the biggest quarterly growth of the cloud computing market.
And it’s easy to understand why: as the global pandemic restricts people’s movements, global cloud infrastructure experiences intense workloads to enable remote working, online learning, electronic commerce, streaming, multiplayer gaming, and virtual presence.
CSPs are gearing up to fulfill these demands by adding more regions. Cloud vendors are shifting to an inclusive strategy to allow more Regions into the developing world. It’s enlightening to note that all three major cloud service providers have at least one Region in every continent except Antarctica.
Gartner’s 2020 Magic Quadrant for Cloud Infrastructure and Platform Services lists Amazon AWS as the market leader, despite Microsoft Azure having the most number of Regions. As this post is written, Microsoft Azure has 39 regions; AWS has 25 Regions, and Google Cloud has 24 Regions.
Microsoft is also the most aggressive in adding new Regions to its already rich Azure Cloud offering. The vendor is coming up with 17 entirely new regions. On the opposite, the market leader Amazon is coming up with 5 new AWS Regions while Google has announced 9 upcoming GCP Regions.
When it comes to Availability Zones (AZs), Google Cloud and Amazon AWS offer at least three AZs in each Region. However, this may or may not be the case with Microsoft Azure.
Microsoft Azure offers three AZs in some regions and points Regions without multiple Availability Zones to the nearest Region with three or more AZs.
Further on, AWS and Azure offer cloud computing to government agencies in the US from dedicated Regions. AWS and Azure have two and seven government clouds respectively. Since these Regions are unavailable to the general public and don’t qualify as global infrastructure, they are excluded from the scope of this post.
Nevertheless, the three CSPs have a presence in all major locations around the world including the US, Europe, India, China, Australia, and the rest of Asia.
The significance of AZs in Cloud Computing
Ideally, cloud service providers (CSPs) offer three or more Availability Zones (AZs) in each Region where services are offered.
AWS Region US East (Northern Virginia), shortened as us-east-1, has six AZs: us-east-1a, us-east-1b, up to us-east-1f. An AZ doesn’t exactly translate to an underlying data center unit and may span more than one data center. Large AZs may contain 4-5 data centers.
While an AZ may span multiple data centers, Availability Zones in a single Region have mutually exclusive data centers. To abstract things, CSPs autonomously plot AZs to identifiers for each account to allocate resources uniformly across AZs in a region.
AZs allow for operating highly available, fault-resistant, and extremely scalable production applications and databases. A single AZ is incapable of attaining such feats. AZs exchange data on high-speed, low-latency networks over completely redundant, dedicated optical fiber networks that allow insane data transfer speed between AZs in a Region.
Encrypted data transfers between AZs over a secure line, while high network throughput enables synchronous data exchange between them.
Availability Zones make it easier to partition applications for high availability. Applications partitioned over AZs in a region are resistant to situations leading to downtimes such as power blackouts, lightning, hurricanes, tremors, floods, etc. A minimum distance separates Availability Zones to ensure they do not share the same environment.
CSPs such as AWS and Azure place a set of AZs at least 100KMs from each other. However, longer distances lead to higher latency and lower transfer speeds.
Single AZ vs Multiple AZs
Most CSPs allow you to host your application on a specific AZ in a Region other than the entire Region. The following illustration depicts a Region with two AZs: AZ-A and AZ-B. However, for certain reasons, the application is hosted only on AZ-A while AZ-B remains unallocated.
The application runs a three-tier server architecture running on a single AZ. The redundant servers may run in each tier but the data center itself is at a sole point of failure.
The following illustration is an improvement over the above architecture. It spans an application over two AZs: AZ-A and AZ-B, which is generally considered a better practice. By replicating all the server tiers in both the AZs: web, application, and the database server, the application will now have multiple points of failure.
The illustration below mimics an AWS EC2 instance and employs Amazon Elastic Load Balancers at each tier. Both Microsoft Azure and Google Cloud offer similar load balancing services. Load balancers route traffic to the available AZs if one or more AZs go down. CSPs, such as AWS, keep load balancers outside of AZs to make them immune from AZ failures.
The ability to employ multiple AZs to build highly available, fault-resistant applications is the foundation of modern cloud computing practices.
The Global Infrastructure of AWS
According to the earlier mentioned 2020 Gartner Magic Quadrant for Cloud Infrastructure and Platform Services, AWS is a strong performer in most use cases. Besides, AWS is a viable applicant for hybrid cloud computing and use cases of edge computing. This cloud provider is the ultimate market leader that continues to grow its share.
- AWS has Regions in all key global population centers
- AWS is the leader in IaaS and PaaS market
- AWS can deliver end-to-end solutions to its customer to demand
- AWS has a strong parent and generated 50% income for Amazon
- AWS and the open-source community don’t go hand-in-hand
- AWS gaining monopoly is a real threat
- AWS is inconsistent over its ever-growing set of services
- AWS doesn’t reduce the prices to its most-used services like EC2 and S3
The Global Infrastructure of Microsoft Azure
Going back to the Gartner Magic Quadrant, Microsoft is referred to as strong in all use cases, including cloud and edge computing use cases, where other CSPs usually struggle. Azure is the go-to choice for organizations invested in Microsoft technology.
It’s fair to argue that Microsoft actively aims to make Azure the center for enterprise applications.
- Microsoft Azure has partnerships with Oracle, SAP, and VMWare
- Microsoft Azure has good standing with the open-source community, unlike AWS
- Traditional enterprises prefer Azure for Microsoft efforts in enterprise computing
- Azure has a particularly strong mind share with Gartner’s enterprise clients in Europe
- Azure has the highest number of Regions of all three CSPs
- Azure has the least number of AZs per region out of all other CSPs
- Azure’s AZs support is limited to Azure services
- Microsoft doesn’t guarantee any capacity despite prepaid arrangements
- According to some users, Microsoft Azure technical support leaves a lot to desire in terms of quality and price
The Global Infrastructure of Google Cloud Platform
According to the 2020 Gartner Magic Quadrant for Cloud Infrastructure and Platform Services, Google has come a long way broadening its strengths and addressing its limitations. It currently provides a strong offering in every use case except edge computing.
Furthermore, Google aims to build hybrid capabilities and affiliations with telecom partners.
- Google is an open-source champion
- Google Cloud experienced noticeable growth in IaaS and PaaS market
- Google Cloud is closing gaps faster with Azure and AWS than anticipated
- Google Cloud is making strides in hybrid and multi-cloud deployments
- Google Cloud is preferred for Big Data capabilities
- Google Cloud lacks the enterprise oomph
- Google Cloud doesn’t make much difference to Google’s revenues
- Google Cloud experienced multiple outages in the past few years
So, the BIG question: Which CSP is a global infrastructure leader?
Amazon’s incumbency may have put AWS in some undesirable positions. Skirmishes with the open-source software (OSS) community and accusations of monopolistic policies were some of past unfortunate incidents. Yet Amazon AWS continues to boost its position as the market leader and is still the go-to choice for most cloud use cases.
AWS has Regions in all major population centers with many more planned to go live soon. Besides, all of its 23 Regions contain at least three AZs. Moreover, AWS has various edge locations to boost its content delivery network (CDN).
Lately, Microsoft Azure is positioning itself as the niche enterprise player in the cloud space with its growing number of enterprise partnerships. Microsoft Azure may have made some real strides in the enterprise market, especially among traditional players; yet AWS is still the dominating force.
Indeed, organizations deeply invested in Microsoft technologies tend to incline more towards Azure over AWS, but even that is not always the case. In most scenarios, Microsoft provides little to no point of differentiation over other major service providers such as AWS and Google Cloud.
Moreover, Microsoft Azure has a deep dive in government cloud contracts. It has 7 Regions in the US dedicated to enabling government services. AWS on the other hand has 2.
Microsoft’s twice the number of Regions over the other two cloud vendors doesn’t put a lasting effect as the penetration of AZs is a little short and lowest in comparative ratio. Microsoft Azure may have 39 Regions to boast about but lacks AZs to balance out the advantage.
Google Cloud Platform or GCP is often considered the distant third among CSPs. GCP grew in prominence over its insistence on developers over enterprises, especially open-source developers. GCP has a comparable number of Regions and AZs to AWS.
As a developer-centric cloud platform, GCP is more flexible than the other two players and is fine with progressive cloud ideas such as multi-cloud and hybrid cloud deployment.
Ultimately, choosing the right vendor for your business will depend on your specific needs and goals. And when in doubt, using the help of an expert cloud computing consultancy can certainly help you alleviate the risks and ensure a successful outcome in your cloud migration journey.
Clouve’s online IT Helpdesk Software allows businesses to efficiently deploy their IT departments in the cloud, without having to worry about unpredicted surges in costs. Clouve’s subscription-based model is designed to help companies with all of their cloud computing and IT-related needs for just a fixed monthly price, no commitment!
Experience the power of Clouve by signing up now, and get your 90-days free trial!