Hyperconvergence is one of the hottest trends in IT infrastructure. Earlier this year, researchers at Gartner predicted, "The market for hyperconverged integrated systems (HCIS) will grow 79 percent to reach almost $2 billion in 2016, propelling it toward mainstream use in the next five years." They added that the market should grow to $5 billion by 2019.
Supporting that forecast, a survey from 451 Research found that among the 32 percent of enterprises planning a major server and storage refresh this year, 86 percent expected to increase their investment in hyperconverged infrastructure.
Given the growing popularity of hyperconvergence, it should come as no surprise that a large number of vendors offer hyperconverged systems. In its Forrester Wave: Hyperconverged Infrastruture (HCI), Q3, 2016 report, Forrester named three pure-play vendors — Nutanix, Pivot3 and Simplivity — as "leaders" in the market. "Strong performers" included big-name IT vendors like Dell EMC, VMware, HPE, Cisco and Huawei, as well as smaller players like Atlantis Computing, HyperGrid (formerly Gridstore) and Stratoscale. And the report noted, "We expect hyperconverged systems to become ubiquitous, fueling a rush to market by all technology vendors."
However, it also cautioned companies to be careful when purchasing hyperconverged systems because of the current marketing hype surrounding the technology. Just because a vendor labels hardware as hyperconverged, that doesn't mean that the product will really deliver all the benefits that organizations are trying to achieve.
What Is Hyperconvergence?
As with any emerging technology, there's some disagreement about the definition of hyperconvergence. Gartner defines it as "a platform offering shared compute and storage resources, based on software-defined storage, software-defined compute, commodity hardware and a unified management interface."
Forrester says that true hyperconverged infrastructure has four key characteristics: integrated compute and storage resources, software-defined storage, automatic discovery and configuration, and minimal management outside the hypervisor console.
Most definitions of hyperconvergence agree on a few key points. First, these are commodity systems that have both compute and storage capabilities. Second, hyperconverged systems include an element of software-defined infrastructure, and third, they simplify management.
This combination of characteristics offers several advantages for organizations.
Vendors that sell hyperconverged systems tout of number of benefits for this style of infrastructure, including the following:
- Ease of deployment — In many cases, deploying a hyperconverged system is as easy as plugging it in and turning it on. The customer doesn't have to worry about making sure the compute and storage components will work together because the vendor has already done the hard work of integrating the system. It also comes with the software-defined storage and management software pre-installed, which can save a tremendous amount of time for IT teams.
- Low cost — Because hyperconverged systems are based on commodity hardware, they should theoretically be less expensive than traditional servers and storage arrays. In addition, the simplified management capabilities and the small footprint for the hardware should contribute to lower total cost of ownership. In reality, however, customers should do their homework carefully because some customers have complained that the promised savings haven't materialized.
- Agility – Hyperconvergence is ideal for virtualized environments. The software-defined storage (and possibly software-defined compute) capabilities make it easy to move workloads around and assign more resources to particular workloads as necessary.
- Scalability — Hyperconverged infrastructure is very easy to scale out. More than one vendor describes their hardware as "Lego-like." As customers need more resources, they simply add more boxes, and they automatically get both compute and storage capabilities as they expand their infrastructure.
- Ease of staffing — Because these systems are easy to deploy and easy to manage, organizations don't need a large IT staff to handle them. This has made them popular with some smaller companies or organizations that are looking for IT infrastructure for remote office or branch office (ROBO) settings.
Despite all these benefits, consultants warn that hyperconvergence isn't right for every situation. Customers should be aware that the technology also has some potential drawbacks.
- Performance — Hyperconverged systems don't always have the most up-to-date components, and therefore, they don't always provide the same level of performance as traditional servers. Analyst and columnist Rob Enderle explained, "This is because part of creating a hyper-converged solution is massive interoperability testing, which can take months to complete after a new processor and chipset are announced." The tradeoff for an easy-to-deploy system is a slower system, which may make hyperconvergence less than ideal for some applications.
- High costs — As mentioned above, customers say that hyperconverged systems don't always save them money. Although they are based on commodity hardware, many vendors charge a premium to support these systems. That's understandable, in part, because the integrated nature of these systems makes it more difficult to troubleshoot whether issues are arising from the compute components, the storage components or the software. Buyers should do a very thorough cost analysis and speak with other customers before deciding on a particular system or a particular vendor.
- Inflexible scaling — The scalability that makes hyperconverged infrastructure ideal for some workloads also makes it less than great for others. If, for example, you have a very data-intensive application, you may need more storage capacity than compute capacity. With hyperconverged systems, you generally don't have the option of adding more storage without adding more compute. That can drive up costs and leave customers with resources that they are not utilizing fully.
- Vendor lock-in — It seems like hyperconverged systems would make vendor lock-in less of a problem — after all, they are based on commodity hardware. In reality, however, once you get started deploying hyperconverged infrastructure from one vendor, it is very difficult to switch to a different vendor without ripping out and replacing your existing hardware. The software that runs and manages these systems is tightly integrated with the hardware, making it hard to mix and match systems from different vendors.
Best Use Cases for Hyperconvergence
Given these pros and cons for hyperconvergence, the technology makes more sense for some use cases than for others. Hyperconverged infrastructure will likely be a good choice for you if you have one of the following needs:
Virtual desktops— Scale-out workloads like virtual desktop infrastructure (VDI) are ideal for hyperconverged infrastructure. Hyperconverged systems easily scale out as additional users are added, and the software-defined environment is ideal for this sort of virtualization. Analysts say that hyperconverged infrastructure that includes SSDs with built-in deduplication are often particularly beneficial for VDI.
Remote office/branch office (ROBO) — As edge computing becomes more popular, many organizations are deploying hyperconverged systems in ROBO environments. Remote locations often don't have on-site IT staff, so easy-to-deploy and easy-to-manage systems are highly desirable in these settings.
Hybrid cloud – Many large organizations use public cloud computing services for many of their workloads, but because of concerns about compliance, security or other issues, they also need to host some workloads in private clouds in their data centers. Here again, hyperconvergence's agility, scalability and support for virtualization make it a very good fit.
Ultimately, whether or not hyperconvergence is right for your situation is going to depend on a large number of different factors. However, most experts seem to agree that the technology is worth investigating.