Monday, December 9, 2024

What is Hyperconvergence? Converged vs Hyperconverged

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

If you’ve been shopping for data center hardware recently, you’ve almost certainly seen systems labeled as “hyperconverged” or “hyper converged.” Other vendors are fond of acronyms like HCI (for hyper converged infrastructure) or HCIS (for hyper converged integrated systems), which refer to the same types of systems.

According to Gartner, sales of hyperconverged integrated systems will likely increase 55 percent in 2018 to be worth about $4.4 billion. And strong growth should continue into 2019, when revenues could exceed $6.4 billion.

The Interop ITX 2018 State of Infrastructure survey found that 65 percent of respondents were either using, planning to use, or looking into converged or hyperconverged systems. Similarly, a Datacore survey found that only a third of enterprises surveyed were not considering hyperconverged systems.

That begs a couple of questions: What are hyperconverged systems? And why are so many enterprises interested in purchasing them?

What Is Hyperconvergence?

As is usual for an emerging technology, the IT industry has several competing definitions of hyperconverged infrastructure.

Gartner defines HCIS as “a platform offering shared compute and storage resources, based on software-defined storage, software-defined compute, commodity hardware and a unified management interface.” It further notes, “Hyperconverged systems deliver their main value through software tools, commoditizing the underlying hardware.”

Enterprise users seem to agree with that definition. When the DataCore survey asked IT decision makers to define hyperconverged, the most popular response was that it included systems that were “tightly integrated with hypervisor but hardware agnostic.”

The most important parts of the hyperconverged definition center on three elements. First, hyperconverged systems unify storage and compute (and sometimes also networking) with a unified management interface. Second, HCI systems are software-defined and rely heavily on virtualization. Third, hyperconverged systems rely on industry-standard x86 hardware rather than proprietary systems.

Within that definition, vendors offer solutions with a lot of variation. Some have tightly-integrated all-in-one hyperconverged appliances. Others offer software-only solutions that support a wide variety of hardware, and still others offer reference architecture for creating your own HCI.

Converged vs. Hyperconverged

One common point of confusion in regards to HCI is the difference between converged vs. hyperconverged systems. These days, there is some overlap between the two categories, but in general, the terms refer to two distinct types of infrastructure.

Converged systems came first. These were essentially hardware solutions that included pre-configured compute and storage capabilities (and in some cases also networking). Some of the solutions also include orchestration or management software, but converged systems are not fully software-defined. As enterprises’ needs grow, they can scale up their converged systems by adding more compute or storage capacity independent of each other.

In hyperconverged systems, by contrast, the emphasis is on the software rather than the hardware. HCI is fully software-defined and virtualizes storage and compute resources so they can be managed as a single pool. And in general, hyperconverged infrastructure scales out, rather than up; that is, enterprises that need more resources add more nodes that include both storage and compute capabilities rather than adding more capacity to an existing node.

A lot of converged infrastructure is based on proprietary hardware, while hyperconverged systems use industry-standard hardware. And the two types of systems are best suited to different use cases: converged for private clouds and greenfield deployments in enterprise data centers, and hyperconverged for creating public cloud or hybrid clouds, virtualized desktops, databases and big data analytics.

Converged Infrastructure

Hyperconverged Infrastructure

Hardware-driven

Software-defined

Scale up

Scale out

Usually proprietary hardware

Usually commodity hardware

Best for greenfield deployments, private clouds and supporting enterprise applications

Best for public and hybrid clouds, virtualized workloads, big data analytics

Advantages of Hyperconvergence

As with most technologies, there are both pros and cons of hyperconvergence. The benefits of hyperconvergence include the following:

  • Simplified management: In the DataCore survey, respondents’ number one reason for choosing hyperconverged systems was to simplify management. The unified management interface makes it easier to monitor and maintain infrastructure, potentially freeing up IT staff for other tasks.
  • Easy deployment: Hyperconverged systems sold as all-in-one appliances are particularly easy to deploy. In most cases, users can simply plug them in, connect them to the network and start using them.
  • Scalability: The second most popular reason for using hyper converged systems, according to the DataCore survey, is that they are easy to scale out. Adding more nodes is usually as easy as (or easier than) the initial deployment.
  • Agility:The virtualized nature of HCI gives enterprises more flexibility. It moves them away from siloed infrastructure and allows them to assign more compute or storage resources to different applications and use cases as business needs change.
  • Low cost: Typically, HCI systems also cost less than other types of infrastructure because they are based on industry-standard hardware. However, enterprises should also consider the potential for vendor lock-in and the training necessary to get existing staff up to speed with the new hardware when calculating total cost of ownership.

Disadvantages of Hyperconvergence

Hyperconverged infrastructure also presents some challenges, including the following:

  • Lack of customization: Because hyperconverged infrastructure is highly integrated, buyers generally don’t have as many options as they would if they were installing separate storage and compute.
  • Inflexible scaling: Because hyperconverged systems scale out by adding more nodes, you generally cannot add just more storage or just more compute power. You usually have to add nodes with both storage and compute, which may not be right for your needs.
  • Potential for high costs: While hyperconverged systems often promise lower upfront costs and lower total cost of ownership than separate servers and storage, that isn’t always the case. Some customers have found that their expenses were higher than anticipated.
  • Lack of hardware reuse: If you already have a large investment in relatively new servers and storage, you may not be able to use that hardware to create a hyperconverged system. It may be more cost effective to wait out your regular refresh cycle to deploy hyperconvergence on a wide scale.
  • Upgrade challenges: Some hyperconverged systems are easier to upgrade than others. Be sure to ask potential vendors whether you’ll need to rip and replace your existing infrastructure when it comes time to upgrade your hyperconverged systems in the future.
  • Performance: In the early days of hyperconverged systems, the available solutions didn’t feature the most recent hardware, and in some cases, they could not provide the performance that end users needed. Vendors have made significant improvements in this area, and some now advertise HCI systems for high-performance computing. Still, be sure to test any system you are considering before buying.
  • Vendor lock-in: Because hyperconverged systems are so tightly integrated, you generally can’t buy servers from one vendor, storage from a second and software from a third. You’ll probably need to source the entire solution from a single supplier, and that could leave you overly dependent on one vendor.

Hyperconvergence Use Cases

Experts also caution that HCIS might not be right for every use case. Other enterprises have found them to be the most suitable for some of the following situations:

  • Software-defined data centers (SDDCs): If you are creating an SDDC hyperconverged infrastructure is a natural choice. Because HCI systems rely on virtualized servers and software-defined storage, they put the environment on track toward a fully software-defined data center.
  • Virtual desktop infrastructure (VDI): In the early days of hyperconvergence, most vendors advertised HCI systems as a good option for virtual desktops, and VDI remains one of the primary use cases for the technology. In Interop ITX survey, VDI was the number one use case for HCIS, cited by 32 percent of respondents, and in the DataCore survey, 27 percent of respondents were using hyper converged systems to support virtual desktops.
  • Databases: As big data stores continue to expand, databases have emerged as one of the primary use cases for hyperconvergence. In fact, in the DataCore survey, databases were the top use case cited by respondents who were using HCI.
  • Remote office/branch office (ROBO) : In settings where organizations have no IT staff on site, such as franchise locations, retail outlets or other remote offices, hyperconverged systems are a good fit because of their simplified deployment and easy remote management. In addition, hyperconverged systems sold as all-in-one appliances often offer a small physical footprint, which makes them even more attractive for these environments. The DataCore survey found that 15 percent of respondents were using HCI for ROBO settings.
  • Edge computing: As enterprises expand their Internet of Things (IoT) deployments, more computing is being pushed to the edge of networks. The same features that make HCI attractive for ROBO also make the systems a good fit for these edge computing situations.
  • Hybrid cloud:When enterprises have virtualized workloads that they are moving between public and private clouds, the software-defined infrastructure of hyperconverged infrastructure is a natural fit. And many cloud computing providers also use HCI in their data centers.
  • Big data analytics: “Bursty” workloads like analytics are also popular for hyperconverged infrastructure. The virtualization capabilities make it easier to allocate more compute resources to analytics jobs as they are needed and then free them up for other uses.
  • Application development environments: For the same reason, hyperconverged infrastructure is also popular for use in dev and test environments. Developers can utilize the compute and storage resources for a short period of time and then return them to the available pool when they are no longer needed.
  • Enterprise applications: Twenty-eight percent of the people surveyed by DataCore said that they were using hyperconvergence for enterprise resource planning (ERP), customer relationship management (CRM), collaboration and other similar workloads. Some organizations are also using HCI for basic file and print sharing.
  • High-performance computing (HPC): Although not all hyper converged systems are capable of high performance, a significant number of enterprises have begun investigating HCIS for HPC needs. In the Interop ITX survey, 28 percent of respondents cited HPC as an ideal use case for hyperconvergence.

Hyperconverged Infrastructure Vendors

Most of the major IT infrastructure vendors now offer hyperconverged products, and the market also includes a few smaller firms that specialized in hyperconvergence. Nutanix is the undisputed market leader, and Gartner also listed Dell EMC, VMware and HPE as “Leaders” in its first Hyperconvergence Magic Quadrant report. Noteworthy vendors include the following (listed in alphabetical order):

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles