Hyper-converged infrastructure (HCI) promises simpler, more flexible and potentially cheaper IT. By combining storage and compute resources – often tightly coupled with virtualisation – hyper-converged brings key resources into a single system.
By removing the need for external storage and networking, HCI should be more reliable and easier to manage. IT departments can either buy integrated HCI hardware nodes or, as is increasingly the case, opt for software-based products.
Although HCI might not match the raw performance of fully optimised, best-of-breed systems, it promises to be quicker to deploy and so more agile and easier to scale.
However, CIOs need to weigh up the benefits of HCI against the potential loss of flexibility and performance, as well as supplier lock-in, which will depend on the use case for HCI. So, what are some of the key benefits?
Benefits of HCI
There are multiple reasons organisations might opt for HCI. These reasons will, however, vary from business to business, and even within an organisation: Hyper-converged could be appropriate for local or branch office deployment, but not be the best choice for the main datacentre.
Ease of deployment
HCI is sometimes described as “turnkey” technology. That might be overstating matters, but easier deployment is a large part of HCI’s appeal.
As the key technology elements – server, storage, and potentially networking and virtualisation – are pre-installed and pre-configured by the supplier, an HCI system is easier to deploy than separate technologies.
This will suit remote or branch office location, with no local IT staff. It is also an attractive model for small and medium-sized enterprises (SMEs). In both cases, HCI systems will ship with remote management, so the central IT team or reseller can fine-tune settings.
Ease of deployment also appeals to startups and other fast-growing organisations, where being able to roll out or scale up IT matters more than squeezing out the last drop of performance.
This is especially important for hyper-converged storage, says Bryan Betts, of industry analysts Freeform Dynamics. Storage is the leading use case for HCI, according to the firm’s research.
“If you want the benefits of a SAN [storage area network] without the costs and complexity of actually building one, use a relatively inexpensive HCI-based virtual SAN instead,” says Betts.
Flexibility
Analyst house Forrester says its clients cite flexibility as the most important reason for using hyper-converged technology.
HCI is flexible, both in how it’s deployed and how it is purchased. Businesses can buy HCI as self-contained systems from suppliers, build their own or even rely on software-defined systems.
According to Naveen Chhabra, a senior analyst covering infrastructure and operations, CIOs are moving away from the DIY approach to HCI in favour of standardised systems and appliances.
“You can buy parts and build an HCI only if you are an engineering-heavy IT organisation,” he says. “Hence you see suppliers that offer bundled appliances are gaining significant market share.”
New purchasing models have added to HCI’s appeal too. HCI systems can form part of hybrid cloud infrastructure, or even – as it is essentially software – run entirely in the cloud.
And on-premise suppliers are starting to move towards pay-per-use models for their hardware, leading with hyper-converged. Chhabra cites HPE’s GreenLake and Dell’s Project Apex as examples.
The number of options open to organisations buying HCI largely offsets the reduced hardware choices within an HCI system, as well as potentially supplier lock-in. “I’d argue that every new revolutionary technology or architecture comes with a lock-in,” says Chhabra.
Software-defined hyper-converged infrastructure reduces the risks from supplier lock-in, as software-based systems can run on multiple suppliers’ hardware.
Lower cost
CIOs can be forgiven for treating supplier claims of lower costs with caution. Nonetheless, public case studies suggest total cost of ownership reductions of 30% and some 20% for operational expenditure (opex) and capital expenditure (capex).
The cost savings from HCI come from two main areas: single sourcing of hardware components – including replacing some hardware elements with software-defined architecture – and from lower setup and management overheads.
Although, on paper, a mixed-supplier approach could lead to a better performing system at a lower cost than one supplier’s offering, the added expenses of integration all too often wipe these savings out, especially for smaller systems deployed to local offices or by SMEs.
Reliability
Reliability is the calling card for single-supplier HCI. Buying off-the-shelf pushes the integration back to the manufacturer. This is all the more so for appliances. Combine HCI with data backup in the cloud, and replacing a defective hyper-converged unit in a remote office could be as simple as connecting the power and local area network (LAN).
In-house developed HCI might not hit the same reliability, as systems will not be pre-tested. But because hyper-converged is based around nodes and clusters, IT teams can design inherently more resilient systems, however they source their hardware.
Unified management, and automation
Unified management, and with that, ease of operational management, is where HCI really comes to the fore.
“Unified management is what marks out HCI,” says Freeform Dynamic’s Betts. “It’s possible with converged systems, but it pretty much defines HCI. The fact that compute, storage and networking is all software-defined or software-driven means it can be run as a whole, as a single entity.”
Unified management promises to reduce the number of tools needed to run IT infrastructure, and it should cut training requirements – IT teams will no longer need to learn multiple suppliers’ technologies and management applications.
Automation is also easier because HCI is homogenous. As the system is software-defined, it is easier to automate tasks such as provisioning and de-provisioning, according to Betts.
Automation, along with unified management, makes HCI easier to scale. A business can add more nodes, or reallocate resources within a cluster, on the fly. Setting up systems in a new office, or even a new territory, could be as simple as delivering the appliances and cloning the configuration.
Unified management also applies to hybrid environments. Again, as HCI is software-defined, the same tools can manage on-premise hardware, hybrid and public cloud. As the business grows, adding more compute capacity or storage is just a question of adding more nodes either locally or in the cloud.
The emerging market for HCI infrastructure on-demand is a further benefit for organisations that need to stand up infrastructure quickly to support innovation.
HCI challenges
HCI, however, does have its challenges. Foremost among these is supplier lock-in and potentially higher costs over time. CIOs also need to assess carefully how much (potential) performance they are sacrificing by buying an integrated system or appliance, against best-of-breed technology.
The key, according to Forrester’s Chhabra, is identifying the right hyper-converged infrastructure supplier –whether that is a software or hardware supplier.
“The heart of HCI is the software, but hardware has its own place in the solution,” he says. “Suppliers would like their clients to believe their respective components matter the most, and this causes a lot of confusion for tech buyers.”
Buyers should also ensure, as far as possible, that suppliers are committed to the HCI path.
And HCI’s flexibility brings its own challenges. CIOs need to decide whether to configure their own hardware, buy appliances outright, focus on software, opt for a rental model, use the cloud, or all of those elements.
With the hyper-converged market still maturing, it is quite possible that organisations will need to use more than one type of HCI to meet their business goals.
