BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Hyper-converged technologies are gaining more attention in the enterprise. The idea of buying a highly engineered system consisting of compute, storage and network in one box appeals to many users looking for simplicity.
But hyper-converged infrastructure (HCI) systems aren't a cure for all IT ills. In a world of virtualization and cloud, using HCI as a building block requires that all available resources are correctly pooled and shared -- which means HCI systems aren't a good fit for all workloads.
For workload optimization, look for specific applications where HCI makes sense. Common examples of these workloads include virtual desktop infrastructure and remote office/branch office configurations.
Target VDI, ROBO with hyper-converged technologies
HCI works well for desktop-serving systems that need virtual desktop infrastructure (VDI). VDI requires a platform that can handle a large number of big-image files and network bursts, while ensuring an optimal end-user experience.
HCI can do a good job of this; it ensures resources are ready to support login storms and minimizes latency across the system, including between compute and storage via internal network connections. This means latency between the user and the system is the only concern.
Remote office/branch office configurations -- provided there are not too many application workloads involved -- can also benefit from an HCI approach. The simplicity of a plug-and-play box means that suppliers can ship HCI systems to site and, as long as someone can plug in an Ethernet cable and connect the main plug, an administrator can complete the setup and monitor and manage the system remotely.
HCI also fits in well with a general virtual machine (VM) strategy. It's easier to deploy VMs to a self-contained platform such as HCI, rather than to a mix of disparate components, for which administrators need to create, manage and provision different settings either via manual scripts or with configuration management tools.
Be cautious with cloud and big data
Hyper-converged technologies can struggle if organizations place too many disparate workloads upon them. This is because optimizing an engineered system to deal with such variability is not easy. HCI can also run into problems if future IT resource requirements are less predictable.
Many HCI systems do not scale out well. If you run out of either compute or storage resources, you may need to buy both -- plus the associated network resources -- just to get the one you need.
Because of these limitations around scaling and disparate workloads, HCI systems should not be viewed as a main component of a cloud platform. Instead, use HCI as a workload-specific platform alongside the rest of a scale-out, cloud platform. If you use it as a main component of the cloud platform, it places your organization over the barrel of vendor proprietary systems, and could demand major forklift upgrades as the HCI system ages.
Finally, online transaction processing and big data are not the best applications to place on hyper-converged technologies for workload optimization. Some specialized systems can be tuned to provide support for these. For example, IBM's PureData, Oracle's Big Data Appliance and others are pre-tuned to deal with data loads, but you would not want to provision any other workload to them. In fact, these offerings tend to be single-workload, hyper-tuned systems for scale-up rather than scale-out.
HCI is a good option for IT teams looking to rapidly acquire and provision a platform for specific types of workloads. Just do not fall into the trap of seeing HCI as a universal answer -- it is only a part of a more complex, overall platform.
Important workload questions for HCI storage vendors
Dell EMC HCI targets new users
Converged vs. hyper-converged for VDI