Traditionally, datacenter technology stacks consist of three discrete layers: compute, storage, and networking. Each of these layers have their own separate hardware appliances and user interfaces (UIs) for management. Hyperconverged infrastructure consolidates these layers within a single standard x86 server or cluster of servers, with a single management interface. By consolidating 3 layers to 1, HCI can reduce the complexity and cost of IT infrastructure.
How Hyperconverged Infrastructure Works
HCI is a marketing term as opposed to a well-defined standard. As a result, there are a variety of different implementations of hyperconverged infrastructure and nailing down exactly what is and what isn’t HCI can be difficult. However, in most cases HCI will converge at least 2 of the 3 layers of the datacenter technology stack.
This convergence that makes HCI possibly comes from purpose-built software running on commodity hardware. The software enables the abstraction of compute, networking, and storage resources. Nutanix, a leading provider of HCI solutions, refers to the layer of HCI that performs this abstraction as the distributed plane.
A management plane sits atop the distributed plane and provides a UI for provisioning and management of hyperconverged infrastructure. This single UI replaces the variety of different UIs IT staff traditionally required to manage hypervisors, virtual machines, storage devices, and network gear.
Converged vs hyperconverged infrastructure
If you’ve kept up with IT buzzwords, you’re likely already familiar with the term “converged infrastructure”. So, what makes hyperconverged infrastructure different? Converged infrastructure refers to a system of components designed to work together, but the components are still discrete. For example, within a converged infrastructure, you may have separate network and compute appliances. With hyperconvergence, the functionality those appliances offer is provided on a single hardware platform.
HCI pros and cons
The promise of hyperconverged infrastructure is to provide more operational agility, scalability, and simplicity when compared to traditional solutions. With HCI, you get the ability to deploy a full general-purpose compute, storage, and networking stack faster and with less management overhead.
This makes HCI stacks useful building blocks for use cases such as micro data centers and other edge computing applications. For example, HCI can also streamline the provisioning and management of on-premises workloads such as virtual desktop infrastructure (VDI) and file storage.
Another upside of hyperconverged infrastructure is that HCI is a good fit for teams focused building a truly software-defined infrastructure. The software-defined approach as a whole adds operational agility and makes it easier to support hybrid cloud environments and lift-and-shift workloads when needed. For example, shifting a workload running on a Docker Container on-premises to an edge container to reduce latency, is simpler than migrating legacy workloads.
Of course, HCI also comes with tradeoffs that make it less than ideal for some use cases. For example, while dealing with a single vendor makes support easier and provides the upside of “one throat to choke”, it also means some level of vendor lock-in is inherent to HCI.
Additionally, while having all the entire general-purpose compute, network, and storage stack is useful for general purpose workloads, it can limit the ability to optimize in certain cases. If a given workload is CPU intensive or requires a large amount of storage capacity, a commodity x86 server simply may not be the right hardware. In short, while general-purpose workloads can benefit from a shift to HCI, it’s currently harder to make a case for specialized workloads.
Example of Hyperconverged Infrastructure
The use of HCI to replace traditional Storage Area Networks (SANs) is one of the more popular examples of hyperconverged infrastructure today. A traditional SAN often contains multiple fiber channel switches, storage arrays, and servers. All these devices have their own discrete UIs and must be racked, powered, cooled, and interconnected using high-speed cabling.
A virtual SAN solution from a vendor such like VMware or StarWind enables to combine the functionality previously provided by the switches, storage arrays, and servers on a single appliance or commodity x86 server. As a result, IT can deploy a SAN with comparable functionality to the traditional model while reducing data center footprint and management complexity. The software-defined nature of HCI can also help simplify creation of security policies, data deduplication, and patching.
- Hyperconverged infrastructure (HCI) is the virtualization of compute, storage, and networking functions on commodity x86 servers
- HCI is a marketing term, so there is variance in what functionality different hyperconverged solutions provide
- A single UI and a software-defined approach to configuration help HCI streamline operational costs and complexity
- Potential downsides of HCI include vendor lock-in and inflexibility for specialized workloads