As AWS CEO Andy Jassy declared from the stage of AWS re:Invent 2017, the cloud is really all about choice. The trick, he says, is that you have to build on a platform with the depth and breadth to actually enable that choice. Though this is obviously self-serving, it also clearly resonates with the burgeoning crowd of developers that want to do more in the cloud.
In his keynote Jassy countered nearly every reason an enterprise might have to not build on the cloud. From IoT analytics to serverless database services to machine learning at the edge, it’s getting hard to find anything that AWS is not building, all while delivering the kind of security and governance that heavily regulated industries require.
But can the company be the “everything store” for enterprise computing without burying the very reason developers first flocked to it, namely convenience?
Would You Like Everything With That?
AWS had so much to announce at re:Invent that it actually started a week early. Once there, analyst Sam Charrington aptly summarized the company’s strategy: “We’ll continue to invest in…basically everything.”
That “everything” included a real-time operating system, new tools to make writing ML and augmented/virtual reality apps easier, impressive new database technology (including Neptune, a fully managed graph database), and more. It was a tour de force of product announcements, the kind of barrage of innovation you’d expect across an industry, not merely one company.
This isn’t just any company, however, and breadth of services is key to AWS’ mission. As Jassy declared, “Customers don’t want to settle for less than half the functionality of the market leader.” ISVs, in turn, also can’t afford to invest in multiple platforms and will pick one default, and maybe another back-up. AWS’ strategy is to be the one-stop shop or, at worst, the first shop an enterprise goes to for its IT needs.
To achieve this, AWS must compete with everyone, including itself.
For example, AWS finally announced a managed Kubernetes service (EKS). Almost immediately, however, Jassy then announced Fargate, “a container-based compute primitive,” such that “All you need to do is build your container image(s), specify the CPU and memory requirements, define your networking and IAM policies, and launch,” as AWS container king Deepak Singh explained.
In other words, you can run EKS to manage Kubernetes, or you can run Fargate and stop thinking about managing infrastructure at all. Or, of course, you can eschew both and opt instead for AWS’ original Elastic Container Service, which sidesteps Kubernetes altogether for container orchestration but is deeply tied into other AWS services that developers use.
But AWS, like its parent Amazon, is the “everything store,” and intends to offer the best options for its customers, whatever their source. As Jassy said, “It’s not about giving people a container service, but rather a choice of container services.”
Another hot area for AWS is the database, and Aurora in particular (“the fastest growing AWS service, ever”). Customers, Jassy noted, are looking to escape the shackles of database giants like Oracle, moving toward open alternatives like MySQL and Postgres. Those same customers, he said, liked the flexibility and pricing of these open source databases but have asked for improvements to performance and scalability, which AWS has engineered into Aurora, DynamoDB, and other services.
In this way, AWS offers helps developers think beyond a one-RDBMS-fits-all model, giving them a bevy of options, some of them built by AWS, some not. In an interview, Singh told me that this guides AWS decisions: what best meets customer needs. If something new is needed, AWS will build it. If a good open source alternative exists, AWS will contribute to it. And if a commercial alternative is available, like the Heptio Quick Start, AWS may elect to simply embrace that.
The key is choice. That may also be the problem.
Too Much Choice?
In closing his remarks, Jassy touted the 1,300 services AWS has launched this year, or roughly 3.5 new feature releases each day. While that sounds great, “choice” can be overwhelming. In the consumer world, for example, retailers exist to make choice manageable. When I walk into a Safeway, for example, I don’t want to be presented with 100 varieties of orange juice. It would simply be too much.
In like manner, there’s a good reason that though enterprise customers demand choice, every product category ultimately filters down to a very small handful of choices, whether in operating systems (Windows or Linux servers), databases (Oracle, SQL Server, or DB2 for proprietary RDBMS; MySQL or Postgres for open source RDBMS; MongoDB or Cassandra for NoSQL; etc.), and so on. Indeed, as pundit Andy Oliver observes, “Choice is often the rationale for a poorly designed complexity trap.”
Along this vein of thinking, it’s conceivable that AWS’ dizzying array of services could end up confusing, not empowering, the very developers AWS hopes to indulge.
If not confusing, as Harvard Business School IT lead Ryan Frazier points out, it’s daunting to try to make use of even a fraction of AWS services. Given that cloud adoption has been driven by developers seeking greater flexibility and, ultimately, convenience, there’s a risk that overloading them with different, sometimes conflicting options may make AWS the “too much of everything store.”
Perhaps, but arguably AWS must innovate like this simply to keep ahead of latent demand. For example, as Hyv’s DL Wilson told me, “(Almost) Everything we at Hyv have been building for… waiting for… Jassy and [Amazon CTO] Werner Vogels are] announcing. All the pieces [we need] to integrate retail and simplify production systems.” In this way, as Vingyani CEO Parthiban Srinivasan puts it, “More functionality plus bare-metal to serverless flexibility is [a] different kind of developer convenience!”
That is, giving developers the ability to do more of their work in the cloud makes their lives more convenient. You fail them if all you deliver is basic compute and storage. Those are just building blocks, but advanced ML/AI, mind-blowing database operations, and more give developers a reason to invest heavily in the cloud and, in particular, on AWS.
This was certainly the message behind Goldman Sachs managing director Roy Joseph’s time on stage. Noting that only AWS inspired Goldman Sachs with confidence that it could execute well enough to match its needs, he said “AWS is like having a safety deposit box in a bank.” Much of that confidence in execution comes down to AWS’ pace and scope of innovation.
So has AWS won? Not yet, though its 44.1% market share, according to Gartner, gives it more than double the market share of all of its competitors…combined. That lead could be lost, however, if AWS doesn’t spend as much time making the depth and breadth of its cloud offerings approachable and, hence, convenient, to developers. So far, so good.
Even so, we’re still in the infancy of the cloud, with increasingly formidable competitors for AWS to beat. Few companies do simplicity better than Microsoft, and Google has been years ahead of AWS in some key areas related to ML and AI.
By introducing competing services, AWS gives enterprises one more reason to stick with AWS as their default, but that default will get called into question more frequently as Microsoft Azure and Google Cloud mature.