Are Your Workloads a Good Match For Hyperconverged Infrastructure?
by Shirish Phatak on July 15, 2017
The pace at which enterprises are adopting Hyperconverged Infrastructure (HCI) continues to accelerate. Gartner estimates that after starting at essentially zero in 2012, HCI will become a $5 billion market by 2019.
What is HCI, and why have so many IT professionals come to love it?
Basically, HCI is an architecture in which compute, storage, and networking resources are tightly integrated into a single appliance. Each appliance functions as a node in a cluster. The key essential of HCI is that all the components of an appliance, as well as the unit as a whole, are controlled at a granular level by software.
Advantages of HCI
Because HCI is fundamentally software-defined, all the intelligence of the system is abstracted out of hardware and into the software layer. Details of the particular hardware components employed are hidden from users behind a unified, “single pane of glass” interface. This means that from a logical perspective, all nodes in an HCI cluster look identical, and the system can scale out simply by adding additional nodes. Endowed with its own dedicated CPU and RAM, each HCI appliance is able to self-manage most of its own configuration and housekeeping functions.
These features of the HCI model vastly simplify the task of configuring and managing a company’s IT infrastructure as a whole, resulting in easier deployment, greater flexibility, and lower costs relative to traditional data centers.
Yet with all its benefits, HCI is still not necessarily the best option for every situation. As the technology matures, it is becoming more adept at running a variety of workloads. But not every workload is a good fit. Before committing to any wholesale adoption of HCI, IT administrators should carefully assess how well the particular workloads they run match up with what HCI does best.
Assessing Your Workloads
In determining whether a particular workload is a good match for HCI, it’s important to look at issues such as latency and IOPS (Input/output Operations Per Second) performance requirements. Said Syed, group manager for hyperconverged product management at Hewlett Packard Enterprise, believes that real-time applications with high IOPS and low latency requirements “don’t make sense” for HCI.
You should also take particular note of how you use virtual machines. According to Clive Longbottom, co-founder and service director at Quocirca, “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.”
An especially important factor to consider is how workloads scale. HCI works best with workloads that scale in a linear fashion, so that requirements for additional CPU, RAM, networking, and storage capacity all grow at pretty much the same rate. Because HCI scales by adding nodes, each of which has a full complement of compute, storage, and networking hardware, a workload that needs more of just one of these can only be provisioned by adding more of all of them, resulting in waste.
Examples of Workloads That Are Well Suited to HCI
The quintessential, and in fact original, HCI workload is Virtual Desktop Infrastructure (VDI).
Says Brien Posey, a Microsoft MVP and former CIO for a chain of hospitals and healthcare facilities, “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.”
Posey also believes that HCI is a good choice for tier 1 workloads, those applications that are essential to a company’s business. He says, “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.”
HCI is particularly valuable for companies with remote and branch office (ROBO) locations. Through the use of HCI, ROBO data can be consolidated in one central location and accessed by users at any site. No longer must each ROBO have someone with a high level of IT competence on its local staff. Instead, administrators at the organization’s main office can monitor and manage the ROBO’s IT infrastructure as if they were on site.
An especially important HCI-friendly workload is backup/restore/disaster recovery. With HCI a sophisticated data protection scheme can be implemented once in software, and applied across an organization’s entire geographically dispersed infrastructure.
Many companies are finding HCI ideal for application development and testing. HCI systems contain their own compute, storage, and networking components that can be entirely segregated from production facilities. Plus they can be quickly deployed with just the number of nodes (as little as two or three) required for the task, then grow by adding nodes as necessary.
Workloads To Be Cautious About
Although workload characteristics are becoming less of a factor as HCI technology matures, many experts still advise caution regarding applications involving the cloud or big data. Citing limitations with respect to scaling and disparate workloads, Quocirca’s Clive Longbottom warns that “HCI systems should not be viewed as a main component of a cloud platform.” He also believes that in terms of workload optimization, “online transaction processing and big data are not the best applications to place on hyper-converged technologies.”
Yet, such concerns are diminishing. Richard Fichera, vice president and principal analyst for infrastructure and operations at Forrester Research, believes that as HCI continues to develop, the characteristics of individual workloads will increasingly become less of an issue. Even now, he says, “these systems can run almost anything except latency-sensitive workloads needing a millisecond or better latency out of the storage.” Adds Enrico Signoretti, Head of Product Strategy at OpenIO, “Hyperconvergence is still a great solution for all those traditional workloads which don’t need large capacity storage or low latency.”
It’s Time To Take a Close Look at HCI
The benefits of HCI are becoming more compelling every day, while its limitations with regard to compatible workloads continue to diminish. Enrico Signoretti sums up the current utility of HCI this way:
“It’s easy to see that in small enterprises (and highly virtualized environments) HCI is the way to go and it can potentially cover almost 100% of all needs. At the same time, with highly specialized applications accessing large amounts of data, such as for Big Data Analytics, we will need different architectures designed with high efficiency in mind.”
If you’d like to explore whether HCI might be a good match for your company’s workloads, Talon can help. Please watch this brief Talon FAST™ video.