Total cost of ownership (TCO) in IT systems is a function of their capacity to do work, which in turn, is a function of storage and compute performance. In a Hyperconverged Infrastructure (HCI), this capacity is expressed in terms of the number of virtual machines (VMs) that can be supported for any given configuration. TCO is also affected by architectural design, as HCIs combine storage and compute resources into a scale-out cluster.
In this TCO analysis we calculate the storage and compute performance of NetApp HCI and two leading competitive systems and compare how their different architectures impact total cost at varying levels of scale.
Benefits of Hyperconverged Infrastructure
HCI technology combines compute power with local storage and storage networking, running one of the available hypervisor platforms in a clustered architecture. A turnkey solution, HCIs have been popular for specific applications that required a new or dedicated infrastructure such as VDI or for isolated environments, like branch offices or remote departments.
As a scale-out solution, HCIs are designed to be deployed quickly, be simple to operate and be easily expanded, making them a good fit for IT generalists or even non-IT personnel, in some cases. These comprehensive infrastructures are flexible enough to support a variety of different applications and can be configured to meet the specific resource requirements of a diverse workload mix – to a certain extent.
Challenges of Hyperconverged Infrastructure
Hyperconverged clusters expand easily, but the mean unit of scale is the node itself. And, often nodes in the cluster must have the same storage configurations. This lack of resource flexibility runs counter to a primary benefit of the hyperconverged infrastructure, the ability to support a wide variety of workloads.
Companies want to consolidate applications, and the infrastructure supporting those applications, into an HCI cluster (one of the findings of a study conducted by Evaluator Group, “HCI in the Enterprise”). This consolidation creates a non-homogeneous collection of workloads that can require a dynamic mix of compute and storage resources. Unfortunately, most HCI clusters force you to add these resources in lock-step with each other, based upon the configuration of the appropriate nodes. But in addition to scaling, there’s a more fundamental challenge in sizing HCI that creates inefficiency and increases cost.