Cloud Predictions for 2017
Prediction time is one of my favorite occasions. I love looking ahead at what trends are taking shape in cloud computing and consumer tech. At the same time, to keep myself honest, I like to take a quick look back at my predictions from the previous year.
For my 2016 predictions, I give myself a B+. Not because I had any big misses but because I wasn't as specific as I would have liked. I talked about some broad trends and I think I nailed every one of them.
Even in a field as fast-changing as digital technology, some trends continue year after year. Back in 2015, I described a burgeoning multi-cloud world, but way back then, I focused on how business, faced with so many choices, would need to look well beyond unit cost when determining value.
The multi-cloud world
Today, multi-cloud is a full-fledged reality, as organizations seek to match each workload to the cloud platform where it will achieve the best performance and cost-efficiency.
Many enterprises today find themselves managing multiple clouds inadvertently and sometime haphazardly, because teams across the business are independently choosing different cloud platforms and providers to best suit their individual needs. I predict that next year we’ll see more companies formulate explicit multi-cloud strategies to best leverage and coordinate multiple cloud providers.
However one gets there, a multi-cloud world comes with a unique set of challenges, including the need for expertise across a larger range of cloud technologies. It also requires managing multiple vendor relationships and more complicated cost tracking.
One of the most critical needs is multi-cloud security — and my predictions for the near term are a mixed bag.
Security’s place in that world
I fear the security landscape will get worse in 2017 before it gets better, given the alleged foreign government involvement in hacking our election, increasingly sophisticated cybercriminals, and the rise of cheap and insecure Internet-connected consumer devices such as digital thermostats and home security cameras (i.e., the Internet of Things).
Everyone should expect that they or their company will fall prey to some type of cybercrime in the near future, whether it's identity or data theft, ransomware or a DDoS attack. I personally was a victim twice last year, and it was a wake-up call that even those of us who consider ourselves relatively security savvy can be targeted. Check out my recommendations to protect yourself and your family from cybercrime.
On the cloud security front, however, I expect positive change. Just a year or two ago, CIOs cited security as one of the top concerns that kept them from moving to the cloud. Increasingly, however, I speak to CIOs who now recognize the foundational level of security that the best cloud providers can deliver as a key benefit of cloud computing.
The additional complexity of multi-cloud, however, needs more than the foundational level of security most cloud companies now offer. I predict companies will rely further on managed service and/or managed security providers to stitch together their multi-cloud security, leveraging both human expertise and ever evolving security technologies.
While security continues to vex, serverless computing is making developers’ lives easier, and if what I’m seeing and hearing holds true, 2017 is the year it will really take off.
“Serverless,” of course, is a misnomer. There will always be computers underneath. Serverless architecture refers to applications that depend on third party services (backend-as-a-service) or custom code (function-as-a-service) — AWS Lambda being the most popular serverless compute service today. What serverless really means is that developers no longer have to worry about managing infrastructure.
As one of our senior software directors told me, “In 2015 my teams spent all our time on containers. In 2016 we figured out that AWS Lambda and API Gateway let us ditch the container management problem. We spent this year building 13 new microservices (REST APIs). You would have to threaten life and limb for me to go back to containers, VMs or bare metal servers.”
Their group is now 95 percent serverless — at greatly reduced cost and headaches.
At AWS re:Invent last month, Amazon introduced a preview of Greengrass, new software that allows connected devices to run Lambda functions, including local compute, messaging and data caching, to run even when the device isn’t connected to the Internet.
This easy serverless setup should help the Internet of Things market expand exponentially, given its ease of use and low cost. And as those barriers to IoT entry continue to fall, we’ll continue to see new players and their devices muscle into what appears to be an almost limitless potential market, which I talk about further below.
The end of Moore’s law
The effects of the end of Moore’s Law (the doubling of computing power every year since the invention of integrated circuits) and the limits of silicon performance will hit more people at home and where they work in 2017.
Machine learning applications that run on public cloud platforms and custom silicon will highlight the challenge. Individuals, companies and market ecosystems will begin to realize: “I can’t do that in my datacenter, on conventional CPUs at an acceptable total cost of ownership — but I can get it in the cloud.”
Meanwhile, competition in alternative silicon will increase and begin killing off some players. The damage will primarily come from companies that design and build CPUs, but it will also begin to affect the future of original equipment and design manufacturers, cloud providers and companies based on appliance-based server-computing products.
If the era of Cloud 1.0 was marked by enterprises testing the waters and figuring out what they could do in the cloud, Cloud 2.0 will see companies wanting to analyze and act on all that data. In the new year, cloud computing will be more visible, and drive a generational split along the same lines as the earlier shift from mainframe to client-server computing, and client-server to web 1.0 to web 2.0 to cloud.
Vital to that shift is machine learning, the science of getting computers to find hidden insights without explicitly being told how to do so.
In the past decade, machine learning has given us the beginnings of self-driving cars, speech recognition and a better understanding of the human genome, among other advances. In 2017, we’ll see it help demystify big data. Machine learning attachments to mainstream data platforms like SQL and NoSQL and even unstructured data will become more common, more visible and more user-friendly.
This will increase the pull for users to shift certain workloads to off-premise clouds, as they seek higher performance, on-demand storage systems.
Also, while specialized processors can make a huge difference in performance and total cost of ownership, many developers, sys-admins and end users don’t know how to use them or can’t afford them. Cloud platforms can provide programming and system administration tools that abstract away the complexity, and make these technologies more available to buy and program.
As cloud continues its domination, is 2017 the year Google’s cloud will not be ignored?
While Amazon continues its world domination of public cloud infrastructure as a service, with Microsoft Azure remaining in a strong and growing second-place position, 2017 may be the year that Google’s cloud firmly stakes a claim as the third-place-and-rising alternative.
After all, Google is home to Kubernetes, which is getting the lion’s share of attention in the container ecosystem. And Google's recent hiring of Sam Ramji, former CEO of the Cloud Foundry Foundation, to serve under Google cloud chief Diane Green, shows it’s serious about appealing to the developers it needs to win over to compete in the cloud IaaS space against AWS and Azure.
On the other hand, Google’s commitment to products outside of its core search business isn’t always as deep and sustained as one might expect. Remember when Google Fiber was going to transform cities’ ability to offer lightning fast and affordable broadband to its residents? In October, Google hit the pause button with little explanation. So while I’m pretty confident 2017 will be Google Cloud’s year, I’m hedging my bets on this one, and will revisit it at the end of the 2017.
What I am confident of, however, is that the Internet of Things will continue its march into our homes and cars.
IoT in your home
More of us will be talking to devices in our home next year. You can’t visit Amazon.com without being bombarded by ads for Echo, the company’s sleek black audio cylinder that’s always connected to the cloud and interacts with you by voice, answering to the name Alexa. And to help accelerate this move to interacting via voice, AWS has opened up the foundational technology that powers Alexa to everyone in the form of a new product called “Lex.” With Lex, developers can create conversational interfaces or bots into almost any product or service.
You can talk to Google Home through Google Assistant, and give commands to your Xbox with Cortana. I don’t think we’ll be waiting long before Microsoft rolls out something similar for the entire home. And as Bloomberg News reported in September, Apple is hard at work on a home device to compete with Echo, based on its Siri voice assistant.
Even your car will be in on the act, with dashboards that mimic your other devices, with always-on wifi connection. Tesla brings its technology to the masses with its launch next year of the more-affordable $35,000 Tesla Model 3. Chevy plans to give the Tesla a run for its money with the Bolt (not to be confused with the Volt), which recently earned a “10Best winner” designation from Car & Driver.
Another trend to watch is the emergence of cars that talk to one another. V2V, or vehicle to vehicle, technology will allow cars to warn one another of conditions up ahead and eventually optimize traffic flow. I think this will start landing in vehicles in 2017 and will accelerate quickly thereafter.
And with the advent of gigabit internet — even as Google Fiber sputters, AT&T is rolling it out in many markets — even more cord cutting should ensue. We’re already at streaming-device saturation; connecting everything makes ever more sense.
All that, and still no jetpacks! Maybe next year.