As 2021 winds down, Amazon Web Services (AWS) is just getting started, with AWS re:Invent kicking off this week. For the past few months, AWS product teams have been beavering away on a variety of updates to existing services and new, as-yet-unannounced services, all hoping they’ll be picked for announcement at re:Invent or, even better, that Adam Selipsky, AWS’ new CEO, will incorporate these new services and features into his Tuesday keynote.
Although AWS is famously tight-lipped about product announcements, it’s actually somewhat simple to discern what the company will announce. As I’ve learned as a longtime observer (and former employee), the company follows certain patterns. This time, it’s a safe bet to take Selipsky at his word and assume the company is going to try to tame its Byzantine service creep (more than 200 services and counting) for customers with vertical industry solutions. Plus it will likely double down on something it announced at re:Invent 2020: significant investments in cloud operations.
Oh, and maybe another billion things. Because this is AWS and it’s not known for restraint.
First, I should stress that I’m not privy to any confidential information. I left AWS in August and was not involved in any service announcements prior to my departure. So, consider this an educated guess based on paying attention to what the company is doing now and has done before—or hasn’t done, as the case may be.
AWS has flirted with industry solutions for some time. If you click the “Solutions” drop-down on its website, you’ll see the company already purports to offer verticalized solutions, except that it doesn’t really. Or, rather, not enough.
Selipsky recently acknowledged to Bloomberg TV that “it doesn’t matter what we did yesterday,” which was almost entirely horizontally oriented infrastructure. The company has started building vertical solutions “for industries like financial services and telco and healthcare and automotive.” Though one can disagree on the AWS approach to customer obsession, that is the company’s guiding principle. Something that was firmly held to be true one year can be discarded the next, depending on customer needs.
As Selipsky said at the Fast Company Innovation Festival, “We have to continue to, over time, dramatically improve the ease of use; come up with more abstractions; come up with higher-level services, like our call center solution, Amazon Connect; find other areas where there are horizontal or industry vertical solutions that our customers are demanding that operate at a higher level.” For those who follow AWS, such candor is unusual. It’s not that former CEO Andy Jassy shaded the truth; he didn’t, but he simply didn’t telegraph where the company was going.
Selipsky has a different way. It’s a sure bet he’ll be announcing increased focus on vertical solutions.
That different way is desperately needed today. The AWS “everything store” can be a chaotic mess of an experience for customers (and developers, as Scott Carey recently wrote). Want to run containers? AWS offers 17 (yes, 17) different ways. Why? The way AWS is structured, each service/product team is incentivized to go build on behalf of customers, even if customers might be better off if they didn’t. The best programmers may delete more lines of code than they write. Likewise, AWS could probably do with some pruning. As in the industry solutions example, customers could benefit from a way to more easily approach and use the varied services that AWS offers.
This is relatively new for AWS. But the “customer obsession” principle is so strong at AWS that eventually the company will do things that it once swore never to do. Remember when AWS raged against the hybrid machine? Now it partners with VMware, delivers things like AWS Outposts, and does more to make it easy for customers to run hybrid architectures. What about its infamous distaste for multicloud? Although the company still dismisses multicloud as wrong for most customers, last re:Invent it rolled out a Kubernetes service that could be run anywhere, including on rival clouds.
In these ways, as well as with vertical solutions, AWS is following not leading. That’s OK. Sometimes, as with Kubernetes (which the company tried for a long time to ignore), it takes time for AWS to listen to its customers. Actually, that’s not quite right. The company is very good at listening to customers, but sometimes it tries to innovate in ways that it feels are better than what the customer says it wants (its Elastic Container Service was meant to tackle the same problem Kubernetes was solving, but better). No matter. Selipsky says AWS is going to strip away the complexity of running its horizontal infrastructure services with a vertical approach. Count on it.
Given past patterns, it’s a safe bet that AWS will double down on its observability story. The company went big at re:Invent 2020 with new initiatives in this area (managed Prometheus, Grafana, and Open Telemetry services), but these were in preview. If AWS stays true to form, it will announce significant upgrades in the areas of observability and resilience. On resilience, AWS gets bonus points because leaders in the space, like Charity Majors, have insisted the fastest path to resilient systems isn’t multicloud.
Beyond observability, AWS will release a new database service (or three). AWS always releases new database services. It will be news if the company does not release a new database service. Actually, if AWS really wants to do something novel, it will stop pitching niche database services (graph, time series, etc.) and instead offer a general-purpose database. The company has beat its “purpose built” database drum for years now, but a quick glance at the industry’s most popular databases shows they’re pretty much all general purpose.
Of course, AWS will announce more than this. It always does. But these feel like reasonable bets, based on Selipsky’s public pronouncements and how the company has operated in the past.
The TL;DR? AWS is about to get serious about user experience, not simply product launches.