Editor’s Note: At the start of this year, Rackspace Technology CTO Tolga Tarhan shared his view on the five tech trends he expected would dominate 2020. We’re closing out 2020, so Tolga is back with an update on how things are turning out and additional insights into the trends we’re experiencing.
Prediction 1: Hybrid 2.0 – The Resurgence
Score: 9 out of 10
This prediction came hot on the heels of Amazon’s re:Invent 2019, where the hyperscaler appeared to finally embrace hybrid cloud with the launch of AWS Outposts. I said I expected AWS Outposts to move the needle for organizations that are ready to be cloudier in their thinking, but couldn’t justify shutting down existing data centers or had hardware that’s not reached end of life.
This has all held true. Customer demand is accelerating for Outposts and Microsoft and Google’s equivalent solutions (Azure Stack Hub and Anthos, respectively). They want what we said they’d want: consistent management platforms and programming models, with the flexibility to deploy to on-premises and public cloud infrastructure.
Yet the prediction features a small but significant miss. The conversation has quickly moved on from talking about private versus public cloud. These solutions have blurred the lines between the two approaches so swiftly and comprehensively that very few care to make that distinction anymore. Now it’s just cloud.
Now, most customers we talk to simply think in terms of choosing whichever of the hyperscalers’ tech stack they want – wherever they want it. Just eight months later, hybrid and multicloud setups are the new normal.
That’s less of a resurgence and more of a revolution.
Prediction 2: Kubernetes wins
Score: 10 out of 10
The headline here was that containers would become a primary enabler for cloud adoption by providing a platform-agnostic way to package and manage apps. And that for containers you can read Kubernetes, for which I said there was no serious competition for deploying infrastructure to support container-based workloads.
And I was right. The Kubernetes project only turned six in June but a Stackrox survey from March found adoption among enterprises to be 86%. It is the container orchestration engine of choice for enterprises. End of story.
The fact that AWS Outposts, Google Anthos and Azure Stack all support Kubernetes is a strong indication that Kubernetes is the way to go for applications running in multicloud environments. And I continue to believe Kubernetes is the best way to migrate existing applications to the cloud, containerizing them on the way in.
It also has the happy side-effect of acting as a forcing function to make organizations better cloud consumers, finally tossing out the bad practices of the past. This was especially true of those organizations who were concerned with “vendor lock in” from the hyperscalers. Containers provides a next-generation environment freeing companies from being relegated to only consuming VM’s, storage, and compute. Containers enable modern patterns and practices in a mostly-portable way.
Prediction 3: Machine learning is gold – the rest are shiny objects
Score: 10 out of 10
Then as now, I could write all day about why speculation about enterprise applications for AR, VR, XR, blockchain and quantum computing is – to put it mildly – distracting baloney. AI and machine learning, on the other hand, have all the hype of these technologies and the killer apps to go with them.
We predicted, and indeed we’ve seen, strong and rapid hyperscaler innovation around pre-made, ready-to-use machine learning services, as competition heats up for these workloads. Wherever they go, the data follows, making this a must-win segment for the top players – and they’ve responded. Witness: AWS SageMaker, Azure Automated ML, MLOps, GCP AI Hub, to name just a few.
Compelling applications for AR, VR, quantum computing and blockchain will likely come along, but right now they are many years away. Meanwhile, machine learning remains a huge opportunity right now: in one survey, the majority (85%) of respondent organizations are evaluating AI or using it in production. Only 15% are not doing anything at all with AI.
Prediction 4: 5G gives IoT wings
Score: 5 out of 10 (for now)
I had high hopes for this one so it stings a little to read the original prediction: “coupled with IoT, the proliferation of 5G in 2020 will usher in virtually ubiquitous connectivity”.
I still strongly believe that 5G will power many transformational IoT applications. But while these network deployments are happening in pockets, things are not moving anywhere near as fast as I had hoped. Disappointingly, no meaningful networks have been deployed and 5G simply won’t be useful until it is ubiquitous. It’s not enough to cover just a few city blocks.
So for now, progress against this prediction is five out 10. However, network providers have a big stake in 5G and they know they’re falling short right now. We’re just over halfway through 2020 and there will be more developments by the end of the year that better back-up this prediction.
Prediction 5: The evolution of the sysadmin
Score: 7 out of 10
I predicted that rather than waiting for a Windows-like cloud management interface to come along, sysadmins and network admins would evolve into more engineering-type roles. There seems little market demand for another abstraction layer that gets between enterprises and cloud native adoption. Instead, this talent will get more comfortable with writing infrastructure as code and continuous delivery and automation.
This one, like 5G, is also happening more slowly than hoped. That’s something I realized when working on a recent customer RFP, which showed just how many IT teams still want to operate infrastructure the same way they have for the past 25 years.
That tells me that, despite the wall-to-wall coverage of cloud and the hyperscalers’ presence in almost every transformation conversation, the industry still has a big education job on its hands.
What would I add?
As noted in prediction one, regarding the resurgence of hybrid and multicloud, the private versus public cloud conversation is becoming an increasingly weird – and incorrect – distinction.
Companies will – indeed, they already have – become more accepting of the fact that they will always have some private cloud. This renders the differentiation obsolete.
In my view, it follows that the most innovative cloud tech will start to enter on-premises data centers way beyond what we see now with Outposts, Arc and Anthos, to include – for example – serverless.
From there, it’s not difficult to also imagine that more and more cloud service models (flexible consumption and contracting, for example) will also transition over. Organizations will start to seek partners and vendors that can build, maintain and operate private data centers — but in a public cloud way.