The increase in the types of convergence for IT infrastructures has led some to wonder whether convergence is becoming stratified. This question has largely been based around the idea that SMB-sized organizations are gravitating toward hyper-convergence infrastructure, while larger, enterprise-class organizations are abandoning hyper-convergence in favor of composable infrastructure.
There is a simple reason for the allure of composable infrastructure in the enterprise: Composable infrastructure enables public cloud-like flexibility on premises.
For those unfamiliar with composable infrastructure, it is essentially a platform for standardizing workload deployment and management. There are three main ways to run enterprise workloads: on physical hardware, in virtual servers and in containers. Historically, IT departments had to manage each of these three workload types separately. Composable infrastructure treats hardware as a pool of resources that administrators can allocate as needed, regardless of whether a workload needs to be physical, virtual or containerized. All of these workload types can be deployed and managed through a common interface.
Composable infrastructure's public cloud-like flexibility has made the technology an ideal choice for the enterprise. Even though composable infrastructure is still relatively new, the technology has attracted a lot of attention because of its ability to deliver business agility. For now, though, composable infrastructure is limited almost excusively to the enterprise because of its cost and complexity.
What about HCI?
Hyper-convergence infrastructure systems have been around for far longer than composable infrastructure and are based around the idea of simplicity through the use of modular, purpose-built components. Hyper-converged infrastructure (HCI) usually consists of a chassis that has been equipped with standardized compute, network and storage resources, as well as a hypervisor and a management layer. All of the various hardware and software components are collectively sold under a single SKU, certified to be fully compatible with one another.
It is this simplicity and modularity that first made HCI so appealing to SMBs. Hyper-convergence infrastructure is designed to be easy to purchase, easy to deploy and easy to scale, and it was easy to get support when needed. All of these factors made hyper-convergence infrastructure a natural fit for the SMB space.
Even so, there is one big issue holding back SMB adoption of HCI: the price. While hyper-convergence infrastructure was once marketed as an inexpensive alternative to enterprise-class servers, the cost of hyper-convergence infrastructure quickly increased to the point that you can easily pay six figures for an HCI system.
Hyper-convergence still has its place in the SMB market, particularly among larger SMBs. Smaller shops, however, are often priced out of purchasing prebuilt HCI appliances. These organizations may opt instead to use lower-cost, open source hyper-convergence infrastructure systems or to simply run virtualized workloads in the public cloud, thereby avoiding the investment in hyper-convergence altogether.
Conversely, although enterprise shops are beginning to deploy composable infrastructure, they have also adopted HCI in recent years. However, the role of hyper-convergence infrastructure in the enterprise is changing.
There is no rule that says only SMB environments can use hyper-convergence or that only enterprise environments can use composable infrastructure. Enterprise environments are likely to continue to use both composable infrastructure and HCI platforms, but for different purposes. While composable infrastructure works well for general-purpose virtualized or containerized workloads, hyper-convergence is better suited -- at least in the enterprise -- as a platform for performing a single purpose or for building private clouds.