Organizations wishing to deploy virtualized workloads have a number of choices. Among the most popular are converged...
infrastructure, hyper-converged infrastructure and the public cloud. Each option has its advantages and disadvantages, so it is important to carefully consider which one will be the best fit for your organization's needs.
A converged infrastructure (CI) consists of compute, network and storage resources integrated into a single chassis. The primary advantage of CI is that the manufacturer certifies the components to be compatible with one another. In many cases, the components included in a CI deployment are also performance-matched. All of this helps to reduce the complexity of deploying virtualization hosts.
A hyper-converged infrastructure (HCI) is based around the integration of compute, storage and virtualization resources into a common chassis. The key difference between HCI and CI is that hyper-converged infrastructure benefits include much tighter integration with virtualization and management software.
Use case showdown: HCI vs. CI
The two environments can have differing use cases. HCI, for instance, is probably best known as a platform for running virtual desktop infrastructures (VDIs). Virtual desktops can be hosted on non-hyper-converged systems, but there are two main hyper-converged infrastructure benefits for VDI environments.
First, VDI vendors, such as VMware, offer hyper-converged systems that are preconfigured and optimized specifically for VDI. Remember that deep software integration is one of the main benefits of a hyper-converged infrastructure. VDI vendors not only preconfigure and optimize the system for use as a VDI platform, but certify the hardware and software to work together.
A second advantage to using hyper-convergence for VDI is that it is easy to increase capacity by adding additional nodes to the deployment.
This modular design comes at a price, however. A hyper-converged system's components can only be used as a cohesive unit. It is typically not possible to upgrade the components within an individual node, or to leverage components individually. In contrast, CI components can be used individually. For example, a CI deployment's compute resources could conceivably be removed and used for other purposes.
Although VDI is probably the most popular use case for HCI, it is not the only one. Hyper-converged systems are designed to be self-contained and remotely managed. That makes them an excellent choice for hosting virtual machines at branch offices. Hyper-converged infrastructure benefits also allow many organizations to employ flexible dev/test environments.
A converged infrastructure is better suited to general-purpose server virtualization. Hyper-converged systems -- though they can be server virtualization platforms -- tend to be better suited for workloads with very predictable resource requirements because they do not usually allow for individual components to be upgraded. This means workload scalability tends to be expensive on hyper-converged systems because scalability can only be achieved by adding nodes. Each node contains compute and storage resources, so an organization that needs to add capacity will have to pay for more compute, even if it's not needed.
While a converged infrastructure tends to be much more flexible, it can still be expensive and time-consuming to scale.
Another option for your workloads
In addition to converged and hyper-converged systems, organizations are increasingly migrating mission-critical workloads to public clouds, such as Amazon Web Services or Microsoft Azure.
There are two main benefits to operating workloads in the public cloud.
- There is no upfront investment in hardware. Public cloud providers simply bill subscribers for the resources they use.
- Workloads can be scaled on an as-needed basis. For all practical purposes, public cloud providers make unlimited hardware resources available to their customers.
But there are disadvantages to using the public cloud, as well.
- The costs can be somewhat difficult to predict. Although public cloud providers give customers tools for estimating costs, the billing formulas are complex, and actual costs can differ significantly from estimates.
- Communicating with the public cloud may require significant bandwidth. This makes the public cloud a poor choice for bandwidth-intensive or latency-sensitive applications. For example, the public cloud would probably be a poor choice for VDI, unless you had plenty of internet bandwidth to support the virtual desktops.
The public cloud tends to be best suited to hosting workloads that need to scale on demand. Many organizations operate workloads on premises, but use cloud-based resources to scale those workloads during periods of high demand.
Hyper-converged infrastructure benefits are not ideally suited to every situation, and neither are the advantages of a converged infrastructure or the public cloud. As such, any hosting platform -- or combination of hosting platforms -- should be matched to an organization's requirements.
More on the advantages of hyper-converged infrastructure with VDI
Handbook: When and when not to use HCI systems
Get the most out of your HCI deployment