Has AWS Peaked?


There can be no real arguing against the fact that Amazon Web Services reigns supreme with regard to public cloud. Its recently announced quarterly results show that AWS is not only gaining revenue, but actually making a “small” surplus. OK, maybe not so small: a tad over half a billion dollars, compared to a $57 million loss for the same quarter in 2015.

What I have found interesting whilst watching it grow is how much like VMware it has become. I can hear you all saying, “It is nothing like VMware.” But please hear me out. AWS’s growth cycle is very similar. Why do I say this?

AWS has become the de facto leader in public cloud in a manner similar to the way VMware dominated the on-premises data center after 2004. Like VMware, AWS has delivered on the early mover advantage. This does not mean that it will continue to dominate, but more on that later.

The AWS re:Invent industry conference is currently seen as the conference to attend. This used to be the case for VMworld. When I ask colleagues in the industry “If you had a free ticket to attend any conference in the world, which would you attend?” they overwhelmingly answer “re:Invent.” These are people who, if you cut them, would have bled purple only a couple of years ago. Take a long look at the list of re:Invent sponsors. Squint your eyes, and you could be looking at a VMworld sponsor list two or three years ago.

AWS, like VMware before it, has nurtured a vibrant ecosystem of partners and symbiotic vendors that piggyback their service offerings off AWS products: vendors like Cloudian, which utilizes S3 and Glacier; Bracket, whose product creates an overlay to bridge the gap between on-premises and public cloud to provide a seamless hybrid experience; CloudVelox, a cloud recovery and data management company; Ravello Systems, which virtualizes the already virtual and provides a templating mechanism to natively run non-AWS workloads on AWS; and evident.io, which analyzes the AWS security state. This is very similar to the ecosystem VMware built and nurtured in the early years, which spawned companies like Veeam, Vizioncore, PHD Virtual, Atlantis Computing, and PernixData, before it started to eat its lunch with product features like vSphere Data Protection and Recovery and Flash Read Cache.

So far so good. It has made AWS a very successful company. However, AWS is now entering a new phase of its development as an entity. Previously, it had had no real competitors. vCloud Air was lacking in features and the investment from Hive Central necessary to drive the regional growth and continual improvement needed to batter down the door of AWS. I view this as VMware’s “Citrix moment”; I consider vCloud Air to be its XenServer purchase. VMware talks public cloud, but it has its feet too firmly planted in its core business of on-premises data centers to make the hard decisions needed to drive its cloud business. To date, AWS has not made this mistake. That is not to say it won’t. There will come a time very soon when its public cloud–only stance will actually be a bane and not a boon. AWS now has a very serious competitor in Azure, Microsoft’s public cloud. Microsoft, however have something AWS can only dream about: a ready-made market. Microsoft also has a serious hybrid play with Azure Pack (2012) and its forthcoming Azure Stack (2016), which allow you to create a local cloud by literally adding a role to a standard Windows server. Hybridity is almost as simple a creating an Active Directory trust.

It is true that AWS has early mover advantage with its cloud. Further, looking at its list of customers, it has a very good revenue stream. Netflix is wholly in the cloud for its streaming business and is arguably AWS’s biggest customer. However, even that is only in single digits. Other customers include Expedia, Slack, and Airbnb. These seem to point to AWS as the new technology company platform of choice. However, there are quite a few old companies hosting on AWS. Pfizer, a global pharma company, for example, uses AWS VPC and EC2 for bursting its data analysis and modeling. Time Inc., a global media company, is moving five of its legacy DCs to AWS. The Dow Jones, another media company, utilizes VPC, CloudWatch, Elastic Load Balancing, and Identity and Access Management.

Although all may seem rosy in the Amazon Web services arena, this is not so, as evidenced by Dropbox’s recent decision to exit AWS and create its own private cloud to save on cost. This is AWSs Achilles’ heel: at scale, AWS is not cheap. Although Netflix, one of its biggest customers, is not actively looking for a second cloud provider, it may be sniffing (per a 2011 interview between Randy Bias and Netflix cloud architect Adrian Cockcroft in which a four-year view is mentioned).

Is This the Peak for AWS?

We at The Virtualization Practice do not feel that this is AWS’s peak moment. Currently there is no competitor that can reach AWS in terms of features, scale (global capacity), or locality (regional sites). However, that is changing rapidly. Google has announced a massive upscale in regional data locations, with ten new regions, two of which are coming online in 2016 (Oregon and Tokyo). Microsoft Azure has rapidly moved from “Hey, we can do that too!” to becoming a serious player in the market. Azure’s improvement in scale and features is frightening. True, it is still playing catch-up with AWS in terms of scale, but it is rapidly catching up on features. Recently released Azure products include Functions, a Lambda competitor, and Microsoft recently announced Azure Cool Storage to compete with AWS Glacier. Microsoft has a massive installed user base, unlike VMware, which heavily miscalculated on vCloud Air in terms of pricing and needs. Microsoft has got a sensible pricing strategy: you only pay for what you use. With Amazon Web Services, you pay even if your machines are powered off; this has led to some nasty surprises for customers. It also results in an overhead in processes, as you cannot just create and forget with AWS: you need to create, use, and destroy. This in turn leads to a serious requirement to re-engineer your applications to be stateless. The Microsoft way is simpler and more akin to the way the VMware worked its magic.

We feel that with Azure, AWS finally has a serious competitor with the business capacity to actually challenge its position of hegemony. Perhaps not today, but in two to three years, the landscape will be significantly different. Ironically, this is the same for VMware, as with the release of Windows Server 2016, Hyper-V has finally come of age, too.

Posted in IT as a Service, SDDC & Hybrid CloudTagged , ,