With the public cloud, presence is a big deal. The number and location of cloud data centers has a major impact on latency -- and because regulations may vary widely from country to country, global scale demands global buildout. Within Europe, for example, all data processing requires adherence to strict controls over where customer data resides.
A portfolio of geographically dispersed data centers is now table stakes for the big three public cloud providers: Amazon Web Services, Microsoft Azure, and Google Cloud Platform. Note that this is not a competition to build out the greatest capacity -- AWS has a very long lead in that department. This is about a race to meet the public cloud needs of customers wherever in the world they may reside.
Not surprisingly, the size of the United States and its density of customers has encouraged the major providers to create large numbers of "regions" -- that is, geographic areas containing enough data centers to provide robust public cloud service, including redundancy should a data center within a region go offline. Take the two leading public cloud providers: AWS has four U.S. regions and Azure has five U.S. regions, providing roughly equivalent geographical coverage (the two define regions somewhat differently).
In recent weeks, we’ve seen AWS and Azure both announce new and coming regions in the United Kingdom, Germany, India, and South Korea. Asia has seen significant expansion in the past few years, with AWS covering four geographies and Azure now in nine, but all providers seem to be lagging in Europe. AWS had a single region in Ireland for a long time, only recently launching Frankfurt. It’s about time we’re seeing Azure and AWS launch a London region, bringing them to three each.
Titans clash as a new combatant rises
But now the location wars are really heating up. AWS and Azure are clearly reacting to customer demand. Cloud is a core business: It generates billions of dollars in revenue at decent margins for Amazon, and it's central to Microsoft's strategy. Both companies have the resources and scale to deploy new points of presence where they see new demand.
The effect of customer demand can be seen in the activities of AWS and Azure in China. Both have specialized local offerings designed for companies in China that are quite independent from their global product. This has been going on for years, with little visibility to outsiders.
The wild card here is Google, which only recently launched a second U.S. region, taking its grand total to four regions globally. Compare this to AWS’s 11 regions (with India, South Korea, and London announced) and Azure’s 20 regions. Google has no operation in China (although it may be in negotiations to re-engage with that market) and the majority of its data centers are in the United States (plus four in Europe and two in Asia, but only one from each is part of the Cloud Platform). This scope makes sense for Google as an advertising company, but it doesn’t help Google Cloud Platform.
Google has the engineering talent, experience at scale, and capital resources to be in the top tier. Moreover, the company has sent strong signals of its seriousness recently: This month, senior vice president Urs Hölzle predicted that Google's cloud revenue would surpass that of its ad businesses in five years, and last week Google announced legendary VMware co-founder Diane Greene will head up Google Cloud.
Such aggressive moves are understandable because the public cloud will be central to the future of business computing. Broad geographical diversity is important for customers, so they can enjoy the best possible performance for public cloud workloads and stay in compliance with their country's regulatory requirements. Global location is a major frontier in the war to serve tomorrow's enterprise customers.