Cloud-native computing is perhaps the most important trend in enterprise IT today. At its core, cloud-native extends the benefits of cloud computing to the entire IT landscape, including on-premises tech as well as the edge.
The open source container orchestration platform Kubernetes has grabbed the cloud-native flag, with a surprisingly rapid uptake in adoption across the Global 2000. Kubernetes, however, is only part of the cloud-native story.
In reality, it reflects a new paradigm for enterprise computing that is in the process of disrupting many established areas of IT (see my recent article on how cloud-native computing is stateless, codeless, and trustless.)
Admittedly, Kubernetes and its burgeoning ecosystem are still immature, and innovation is proceeding at a rapid pace. Nevertheless, the writing is on the wall: cloud-native computing will disrupt many enterprise IT markets, and with them, the way enterprises implement technology. Here are three such markets.
The End of Infrastructure as Code
The ‘infrastructure as code’ movement was one of the primary technical drivers of DevOps several years ago. Instead of dealing with all aspects of configuring and managing the operational environment, the theory went, we should write scripts or recipes that automate all that configuration for us.
Infrastructure as code in conjunction with cloud computing led to a complete rethink of how to deal with production environments, as the best way to fix a problem in production was often to deprovision and reprovision using an updated recipe, instead of monkeying with individual servers – the now familiar ‘cattle not pets’ approach.
The problem with infrastructure as code is, well, the code bit. As the recipes became increasingly complicated, they required their own lifecycle, including security reviews, testing, and limited deployment. By streamlining one software lifecycle we inadvertently created another to manage.
Cloud-native computing, in contrast, follows the alternative to infrastructure as code: model-driven declarative representations of the operational environment. In essence, create a model of how the environment should behave, and allow the DevOps platform to ensure it conforms to the model. If there’s a problem, update the model.
Cloud-native takes this idea of model-driven declarative configuration and runs with it, applying it across the entire IT landscape, what I call the ‘codeless’ principle of cloud-native computing. As the implementation of this principle progresses, expect to see increasingly sophisticated models that represent the entirety of the enterprise IT landscape – and as a result, infrastructure as code will become a thing of the past.
The Waning of SD-WAN
First generation software-defined wide-area network (SD-WAN) products served two core functions. First, they abstracted the choice of connectivity for WAN links, enabling organizations to choose MPLS, broadband, or other protocols as a matter of policy.
Second, these earlier products shifted the control over remote office networking equipment (which might be at retail locations, factories, etc., as well as offices) to a centralized management facility.
These basic capabilities will remain important in the cloud-native world, but will become aspects of a larger capability: abstracted control and trustless security in general.
‘Trustlessness’ is another cloud-native principle: an update of the now-aging ‘zero-trust’ approach, combining zero-trust and software-defined networking, thus brining the approach forward to the cloud-native computing paradigm.
Trustlessness applies to the network in particular. We must assume that the physical network offers no security protections in itself. Instead, we must create an abstracted, software-defined layer that handles all necessary security functionality independent of the physical characteristics of the network.
First-generation SD-WAN technology has partly achieved this vision of trustless networking, centered on updating or replacing the older notion of a WAN. In the cloud-native world, in contrast, WANs are irrelevant, as we’ve abstracted all network endpoints, regardless of whether they be in a cloud, in an on-premises data center, or on the edge.
This software-defined approach to network control and security is becoming increasingly important as edge computing becomes established in the enterprise – which the explosion of artificial intelligence is driving today, and 5G will continue to drive over the next decade. SD-WAN is in the process of rising to these challenges, and when it does, we’ll probably call it something entirely different.
No Love for Hyperconverged Infrastructure
I’m still confused about why hyperconverged infrastructure (HCI) is even a thing, and I work with a number of HCI vendors.
Converged infrastructure is relatively straightforward. Instead of having separate boxes for compute and storage in your data center (say, because you have a storage-area network, or SAN), combine (or ‘converge’) compute and storage onto the same boxes, making them easier to configure and less expensive in the process.
Hyperconverged infrastructure adds virtualization to the mix – where the hyper prefix comes from hypervisor. With HCI, you have some mix of virtualized compute (traditional hypervisor-based virtualization), virtualized storage, and virtualized network.
The underlying hardware is likely to be converged infrastructure, but there’s no requirement that it must be. You might have some virtual machines, a virtualized SAN, and some software-defined networking similar to how SD-WAN works. That entire mélange might qualify as HCI, if a vendor happens to sell it that way.
This inherent vagueness as to what HCI actually is gives license to vendors to build dramatically different offerings and call them all HCI, leading to a severe case of caveat emptor.
That being said, the principles underlying HCI are sound, and in fact are core to cloud-native computing. Cloud-native infrastructure requires virtualized compute, virtualized storage, and virtualized network – only now, there’s no particular requirement to bundle these three things into a single box so that a vendor can slap a SKU on it. Instead, virtualization simply comes with the cloud-native territory.
The specifics of the implementation of that virtualization will depend on the problem you’re looking to solve. In fact, even though Kubernetes is at the center of the cloud-native whirlwind today, the container virtualization abstraction is but one of several options. In reality, cloud-native computing comprises traditional virtualization, containers, and serverless computing as well.
Virtualized storage is also an essential enabler of the stateless requirement for cloud-native computing. How do we manage state information in an inherently stateless environment, without compromising cloud-native principles? Hint: start with virtualized storage.
And the virtualized network, of course, is the next generation of SD-WAN.
The Intellyx Take
Full disclosure: we number infrastructure as code, SD-WAN, and HCI vendors as our clients. We help them with their marketing strategies and content marketing needs. Is this article biting the hands that feed us?
Not a chance. They hire us in large part to give them the big picture of the disruptions that are likely to impact their businesses – and cloud-native computing is one of the most significant disruptions since the Internet itself.
It’s also important to note that each of the three technologies in this article’s crosshairs are transitional in nature – and as such, are enjoying an extended day in the sun. Each of them met customer needs at the time, and will continue to do so for a while yet.
But as with anything I might call transitional, the end of the status quo is fast approaching. Vendors – as well as the enterprises that depend upon them – are well advised to look over the horizon to plan on the Next Big Thing. And in enterprise IT, the Next Big Thing is cloud-native computing.
Copyright © Intellyx LLC. Intellyx publishes the Agile Digital Transformation Roadmap poster, advises companies on their digital transformation initiatives, and helps vendors communicate their agility stories. As of the time of writing, none of the organizations mentioned in this article are Intellyx customers. Image credit: Michael Coghlan.
