Hyperconvergence: What’s all the hype about?

One of the hottest trends in data center technology is hyperconvergence, with early adopters reaping the benefits of cost savings, enhanced data protection, increased scalability and ease of management.

So, what is hyperconvergence? It’s a way to simplify data center operations and management by combining compute, storage and networking in a single, software-driven appliance.

Gartner predicted the market for hyperconverged integrated systems (HCIS) would grow nearly 80% to reach almost $2 billion in 2016, propelling it toward mainstream use within the next five years. HCIS will be the fastest-growing segment of the overall market for integrated systems, reaching nearly $5 billion (or 24% of the market) by 2019.

“HCIS becomes the top-selling integrated systems form factor by 2019, although more traditional blade/SAN [storage-area network]-based technology will remain viable in the right use cases,” says Andrew Butler, vice president and distinguished analyst at Gartner.

“It’s not all about HCIS, but the market mood is steadily shifting in [its] favor,” Butler says. The market is morphing into a new business model where the HCIS vendor creates the software, then either works with one or more hardware vendors to create a certified design or relies on buyers to provide their own hardware, he says.

“We also now see the big systems vendors entering the market with their own designs,” Butler says. Gartner predicts that the big system vendor market share for HCIS will increase from about 5% today to 60% by the end of 2017. Gartner puts Cisco, HPE, EMC, NetApp, Oracle, Dell and others in this category.

+ ALSO ON NETWORK WORLD Nutanix CEO skewers box-based hyperconvergence rivals +

Since hyperconvergence is often touted as a private cloud or hybrid cloud solution, HCIS vendors are increasingly modeling the pricing and deployment of their solutions on the pricing models used by public cloud services providers such as Amazon.

“HCIS vendors are not trying to compete with Amazon; rather their aim is to enable data centers to deliver similar services to their lines of business,” Butler says. “In the coming years, we expect new generations of composable infrastructure to further evolve the market for HCIS.”

Hyperconvergence “is an outcome of the fact that blade/SAN-based systems are totally unsuited to use at the network edge, Butler says. “So hyperconverged systems always dispense with the SAN, replacing this with a software layer and local disks distributed across the nodes.”

Among the main benefits of the technology is the ability for organizations to start small and grow resources at a very granular level, Butler says.

“So the minimum investment in HCIS can be as low as $20,000 to $30,000, whereas a blade/SAN-based system generally requires an investment of $300,000 or more,” he says.

Because hyperconvergence can be achieved at relatively low cost, it becomes well suited to remote offices and other branch locations, and makes converged infrastructure more affordable for smaller companies, Butler says.

“HCIS is also a good fit for point projects within a larger data center, where organizations may want to deploy an appliance approach for new generation workloads” such as virtual desktop infrastructure (VDI), he says.

While the initial starting price for HCIS can be low, it’s a fallacy to assume it will always be cheap, Butler says. “A mature HCIS project, with perhaps 100 or more nodes, will represent a huge capital cost that could run into the millions,” he says.

The technology is usually easy to set up, and vendors are investing in more and more automation that simplifies the task of initial set-up or adding new resources, Butler adds.

Pitt Ohio turned to SimpliVity

Transportation and supply chain company Pitt Ohio first delved into hyperconvergence at the end of 2014, when it launched a three-year plan to refresh its data center infrastructure. The data center included storage, networking and computing components from multiple vendors.

These were not only costly, but difficult to manage, said Justin Brooks, systems engineer at the company. “We also had been looking at DR [disaster recovery] solutions over the past couple years, but it was always a hard sell to the business because of the added cost,” Brooks says.

“This is when we started looking at hyperconvergence as a possible solution. Once we worked out the numbers and realized the goals we had in mind were obtainable, we took the plunge.”

Pitt Ohio deployed a hyperconvergence platform from SimpliVity into its production environment. This consists of a Citrix environment that hosts desktops and applications throughout the organization, including three subsidiaries.

“We also have some application, file and SQL servers mixed in,” Brooks says. “So far with our first phase deployment of SimpliVity we have replaced eight UCS blades, reduced our [storage] footprint, reduced licensing costs and have been able to decommission one of our SANs.” The company expects to phase out its SAN along with all of its storage appliances when the second and third phases of the implementation are complete.

Among the key benefits Pitt Ohio is seeing or expects to experience are enhanced data protection, disaster recovery, improved data efficiency, scalability and ease of management.

Server backups can now be done in seconds compared with hours. “We have been able to shorten our backup window to one hour on SimpliVity compared to eight hours” with the previous infrastructure, Brook said.

+ MORE ON NETWORK WORLD Simplivity attacks the ‘unbearable complexity’ of IT +

For DR, instead of having multiple pieces of expensive hardware sitting off-site, the company now has a few nodes with top-of-rack switches.

“We have been able to realize more predictable performance for our Citrix environments,” Brooks said. “We have also been able to get more server density, allowing us to replace eight [blades] with three SimpliVity nodes.

The hyperconvergence strategy has proven to be ideal for use cases such as VDI. “The SimpliVity solution not only does compute and storage, it does backup, DR and optimization,” Brooks said. “It does this with no added cost or licensing while being managed from a single pain of glass.”

CarePoint goes with Nutanix platform

Healthcare provider CarePoint Health System in New Jersey, launched a hyperconverged infrastructure about two years ago. The main goals of the project were to simplify IT management, break down information silos and reduce costs, says CTO Lev Goronshteyn.

CarePoint deployed a hyperconverged platform from Nutanix, consolidating workloads that were on blade servers and SANs. “We have migrated most of our physical infrastructure and converted to virtualized workloads on our [hyperconverged platform] HCI as well,” Goronshteyn says. The company is in the process of creating a hybrid IT model, with some workloads running on premises and some in the cloud.

The move to a hyperconverged environment will enable CarePoint to reduce its costs by more than $15 million over a five-year period, Goronshteyn says. “We are now able to do one-click upgrades on all our [IT] environments with no downtime and impact to our end users,” he says.

+ MORE ON NETWORK WORLD VCE chief boasts of hyperconvergence superpower +

The simplicity of the platform “allows us to focus our attention to more important projects rather than focusing on maintaining infrastructure,” Goronshteyn says. “It also reduces the amount of time it takes to deploy any type of workload, which in turn makes our department more efficient by allowing us to keep up with our business needs.”

For some companies, such as marketing services provider Tele Business USA, improved systems availability is a big benefit of hyperconvergence. Tele Business launched into hyperconvergence about three and a half years ago, says Adam Miller, senior network engineer.

“We had many servers that were just single boxes running one or two things,” Miller says. “And we also had a couple of VMware servers, but they were not connected to each other so if one of the servers died we would have to restore from backups. Hyperconvergence seemed the best option for us.”

The company deployed a hyperconvergence platform from Scale Technology, converting more than 90% of its environment.

“Redundancy is the huge benefit of hyperconvergence,” Miller says. “The fact that I could have a hardware failure and lose an entire node and not lose any data is huge.”

With its hyperconvergence setup, Tele Business has created its own private cloud, which Miller says is still a work in progress. “The last few updates on the Scale nodes has now allowed me to replicate from one set of nodes to another without using third-party software,” he says. “This is more cost effective for us than using servers in the cloud that we would need to pay for by the month.”

Deploying hyperconvergence can come with hurdles. “Any integrated system creates organizational challenges,” Butler says. “More established companies with entrenched and hierarchical IT organizations will often struggle to reconcile the politics of integrated systems, which blend multiple IT roles and disciplines.”

In addition, all integrated systems are usually proprietary. “They tie the buyer into a long-term relationship,” Butler says. “So all integrated systems impose a set of rules that do lock people in, but this is a necessary by-product of the ability to deliver highly integrated and reliable management.”

As the technology matures and configurations get larger, many users are questioning the ability of these systems to truly scale, Butler says. “Only now are some, but not all, HCIS vendors investing in technology that will better integrate the role of the switch, and demonstrate that very large clusters of 20 nodes or more do not have to compromise network performance,” he says.

Violino is a freelance writer. He can be reached at bviolino@optonline.net.