Hyper-converged infrastructure has become a popular choice for IT departments because it's easy to deploy and scale....
However, some workloads are better suited for hyper-converged platforms than others.
The best known use case for a hyper-converged platform is supporting a virtual desktop infrastructure (VDI). What makes HCI so well-suited for VDI environments is predictable scalability. Each node can host a certain number of virtual desktops, and because each node contains identical hardware, it's easy for IT administrators to determine the total number of nodes needed to host their desired number of virtual desktops.
Hyper-convergence is also a popular choice for VDI because each node contains its own storage hardware. This eliminates the need for costly SAN storage, and the distributed storage architecture also eliminates performance problems that might otherwise occur as a result of all of an organization's virtual desktops sharing a common storage pool.
Use cases beyond VDI
Another workload suited to HCI is hosting an organization's tier-one applications. Hyper-converged platforms are often built to provide the kind of high availability that companies need for mission-critical workloads. And vendors offer additional layers of redundancy, such as storage redundancy and the ability to mirror entire nodes or clusters.
Hyper-convergence has also proven useful in testing and development environments, which have long posed challenges to enterprise IT. Test/dev environments must be similar to the production environment for tests to accurately reflect the way code will behave. At the same time, it is critically important to keep testing and development environments isolated from the production environment. This prevents the test/dev environment from interfering with production resources.
IT departments often repurpose aging, decommissioned server hardware for use in test/dev environments, but outdated hardware does not make for an optimal environment. Hyper-convergence works very well in these environments because HCI systems contain compute, network and storage resources. The development team can use a hyper-converged platform to establish an entirely self-contained testing environment without relying on outdated hardware. And the modular design of a hyper-converged infrastructure means the development team can purchase only the hardware it needs now with the promises of expanding in the future by simply installing additional nodes.
The modular design of a hyper-converged system and its ease of deployment make it a popular choice for branch offices, too. Many vendors design their hyper-converged platforms to support remote management, so an administrator who works in the main office can manage branch office infrastructure as easily as if it were local.
Workloads to avoid
There are at least a couple use cases companies shouldn't run on HCI, such as applications that don't have balanced system requirements. A hyper-converged platform is a combination of performance-matched compute, network and storage hardware, so if a workload places an abnormally high demand on one of these resources, others may be underutilized. This becomes especially problematic when it's time to scale.
For instance, suppose an organization runs a storage-intensive database application. If the company hosts it on a hyper-converged system, the app might consume storage capacity or storage IOPS to the point that the storage hardware can't accommodate any other workloads. Because storage is tightly coupled with network and compute resources in hyper-converged platforms, the organization will have paid for network and compute resources it can't use.
Hyper-convergence also tends to be a poor choice for organizations that run hypervisors from multiple vendors. Part of the appeal of a hyper-converged platform is that the hypervisor and management software are tightly intertwined with the rest of the system. In most cases, the relationship between hardware and software means that a hyper-converged system will only support a single vendor's hypervisor. Those organizations that must run multiple hypervisors might consider using a more traditional architecture or running parallel hyper-converged platforms -- one for each hypervisor.
How HCI might harm VM performance
Approach hyper-converged options with workload size in mind
Storage workload performance not a sure thing with hyper-converged systems