Hybrid cloud vs. hyper-converged: Hybrid wins hands down
When everything goes to cloud, why do you even need HCI? Who cares where server, storage, virtualization and networking are when IT adopts a 'run any app from anyplace' strategy?
If you are a big IT infrastructure vendor, how do you successfully pivot from traditional data center business to stay relevant in the inevitable new world of cloud and hybrid cloud architectures? You would certainly try to transition your current strengths in infrastructure into strengths in cloud architecture and services, as some hyper-converged vendors have done. How traditional vendors pivot to a cloud/hybrid cloud is especially important to ask in regard to hyper-converged vs. cloud computing, as some vendors have started to use the terms hyper-converged infrastructure and hybrid cloud architecture interchangeably.
The physical data center may never completely go away due to security, compliance and other indirect factors, but with cloud computing, there is no longer a functional mandate to physically centralize all data and compute. The idea of data "center" is blurring into a virtual concept that is more about hosting workloads and data where cost and service are optimized, rather than as a single physical location where a sys admin can go and touch all the machinery.
In making this pivot, I've observed the strategic path taken by many IT vendors is to recast their traditional on-premises technologies as hybrid cloud-building componentry. They do a deal or two with major cloud providers to host virtual versions of core intellectual property, acquire some data migration tools, layer some new remote managed services over the top and then stir it all up with marketing magic.
While at times these transformations are hard to follow, this hybridizing recipe is actually not bad for helping customer organizations evolve. In fact, it can provide smooth continuity in cloud adoption for more careful cloud journeyers in terms of skills, processes and past investments.
Nutanix HCI clusters span private and public clouds. 'But it's not about hyper-converged infrastructure, it's about hybrid cloud infrastructure,' Nutanix CTO Rajiv Mirani in September 2020 explained.
Emerging hybrid cloud innovation
We commonly see two innovations emerge that are worth calling out in the cloud hybridization process.
The first is an abstract fluid workload layer above infrastructure that allows users, their apps and their data to run "anywhere" -- on premises, virtual private cloud, public cloud and even out to the edge without modification. This hybrid cloud fluid layer is fast forming a new field of competition between the likes of VMware, Nutanix and others (including public cloud providers), where each wants to be seen as the best "secret sauce" for enterprise architectures.
The second is remote management services evolving into remote data center operations. When the latter is combined with a financial model that shifts an organization's cost basis from equipment and licensing to computing and storage units used, they are then essentially subscribing to services from their own on-premises dedicated cloud. The IT vendor in essence becomes an on-premises "private" service provider, retaining ownership and providing remote management for any on-premises infrastructure.
Together, these innovations now offer most organizations opportunities to adopt the cloud faster. IT shops should be careful to avoid becoming deeply entrenched with a single vendor, however. Be sure to allow for future "fluidity" between fluid infrastructure layer vendors as much as for future multi-cloud mobility between cloud and service providers.
Hyper-converged vs. cloud: A hyper hybrid cloud?
One part of this observed IT vendor pivot to hybrid cloud stretches a thin argument. More than one vendor has proposed their hyper-converged infrastructure (HCI) makes them a great hybrid cloud fluid layer. Nutanix, for example, even attempts to redefine the HCI acronym as Hybrid Cloud Infrastructure.
Now, I have long been an advocate of HCI and hyper-converged building blocks as a great way to build a scale-out cloud or cloudlike platforms. But once IT services are cloudlike, end users don't (or shouldn't) care about the shape of the underlying actual infrastructure. They consume infrastructure as a utility, either as virtual infrastructure, infrastructure as a service (IaaS) or embedded in higher layer offerings such as platform as a service (PaaS) or software as a service (SaaS). The physical infrastructure becomes solely the service provider's responsibility.
If an IT shop hosts and builds its own on-premises private cloud, HCI would be a good architectural choice. However, when we consider hybrid computing, we are -- by definition -- really 'above' the on-premises infrastructure layer.
If an IT shop hosts and builds its own on-premises private cloud, HCI would be a good architectural choice. However, when we consider hybrid computing, we are -- by definition -- really "above" the on-premises infrastructure layer.
Certainly, virtualization and containerization management services play a big role in the hybrid-enabling fluid layer. Some have even conflated virtualized infrastructure -- and the software-defined resources that run in them -- with hyper-convergence. HCI can be a great platform for virtualization (and virtualized resources). But the real point of core hyper-converged infrastructure derives from its internal marriage of formerly siloed infrastructure into modular physical building blocks.
A virtualized machine doesn't know what infrastructure it's running on, and really shouldn't care if it's currently hosted on hyper-converged infrastructure, a public cloud instance, a dedicated server, a mainframe, a composable machine made up of temporarily assigned resources, or an edge device. That kind of fluidity is really the main benefit of hybrid architecture.
Where are my workloads?
Ultimately, the rationale for hybrid cloud architecture is to enable workload placement optimization. Hybrid approaches are becoming popular because it's hard to balance all workload and cost requirements within only one compute environment.
While there is significant cloud adoption friction keeping the lights on in monolithic enterprise data centers today, many of these concerns will abate as security, compliance and other management tasks become engrained and automated at a unifying cloud hybridizing management layer. Also, as fast networking reliability improves, it will matter less again where your workloads are hosted just for performance reasons.
Up-and-coming technologies such as storage class memory, edge computing and peer-to-peer protocols will further unlock constraints on hybrid cloud architecture.
Eventually, my vision is that apps and data won't be predetermined to live anywhere specifically. Instead, compute and data hosting will become the job of controlling AI/machine learning algorithms that will dynamically plan and deploy functions, data and streams of data across the hybrid landscape to continually optimize cost against service requirements.
Even if this is a distant vision, the hybrid fluid layer will grow intelligent and eventually standardize into a modular "plug and play" control plane. Today, many IT vendors make the argument that they are best positioned to provide the single best hybridizing architecture choice, many by virtue of their traditional data center offerings. It's worth asking if those claims really matter when building a new cloud architecture.
I'd also bet that open source hybrid cloud technologies will emerge and threaten to disrupt one or more of the big vendor hybrid cloud reinventions within the next five years.
Hyper-converged infrastructure can be a great architecture for IT modernization initiatives and accelerating internal digital transformation projects. Hybrid cloud and hyper-converged aren't synonymous, however. The latter is an enabler of the former.
For those still on the fence about taking advantage of hybrid cloud offerings, it's easier than ever to try out new services, as most cloud offerings and managed services are just one click away. Just remember that HCI adoption can contribute to a faster hybrid cloud journey, not the other way around.
