As companies grow their IT systems tend to expand in a way that can lead to them becoming unwieldy and hard to control. Hyperconvergence is about simplifying things by consolidating the IT infrastructure into a virtualised system, this reduces the number of devices, cuts maintenance costs and ensures there are fewer potential points of failure, making systems more resilient.
Essentially it’s a software defined architecture that merges storage, networking and other resources on a single system based on commodity hardware. The benefits of this include being able to manage the whole thing as a single system through a common tool set.
How it works
Hyperconvergence started in the idea of converged infrastructure, where a vendor provides a package of pre-configured software and hardware in order to simplify management and minimise connectivity issues. The key difference is that in a converged infrastructure the technologies can be separated and used independently. In a hyperconverged system they’re so closely linked that they can’t be broken down and used separately.
Hyperconverged infrastructure (HCIS) systems generally use x86 architecture in a dedicated appliance, additional appliance nodes are easily added, so configurations can be scaled to create clusters. This allows tasks to be distributed between clusters so that the system can continue to work if an individual component fails. It also provides a plug-and-play capability making it easy to expand the system with minimal disruption.
In conventional setups where servers, storage and networking are all separate, components can be upgraded individually but this doesn’t increase the performance of the overall system. In a hyperconverged environment adding extra appliance nodes scales up all elements simultaneously, so the performance of the system improves across the board.
Hyperconvergence is closely related to software defined networking and storage. The building blocks of software defined systems are very similar to those used for hyperconverged infrastructure. There are similarities too with hyperscale web services which rely on similar configurations of commodity hardware driven and managed by software to create custom systems.
Who is it for?
You might by now be thinking that hyperconvergence is something that is only for large organisations. In fact it can benefit businesses of almost any size. Smaller enterprises that might have just two or three servers providing, database, storage and mail can suffer a major outage – and loss of custom – if one of those servers fails. By virtualising those functions on a hyperconverged system, even with only a small number of nodes, the business gains greater protection against systems failure.
How easy it is to hyperconverge a business depends to an extent on your existing infrastructure. If all systems are Intel based then it should be relatively straightforward. However, if there’s a mixture of architectures, Power PC and SPARC for example, then converging will be more of a challenge and may require some significant changes.
In moving to a hyperconverged model there are a number of things to consider. One of the big advantages of HCIS is its resilience and ability to tolerate failures, but this needs to be designed in from the beginning.
It’s important to think about all the things which might fail and eliminate what are known as ‘single points of failure’. There is no point in having your virtual architecture on just one physical host for example or you’re no better off than with a conventional setup. Similarly you need to have multiple connections to your network and make sure that systems are protected from power failure.
Planning for expansion is vital too. Although it’s easy to add extra nodes to your HCIS you still have to physically house them in a rack, provide power and so on.
For enterprises that rely heavily on their technology, eCommerce businesses for example, HCIS makes a lot of sense. Wherever there is a single point of failure there’s a risk that your entire operation could grind to a halt. This is something that should concentrate the minds of management when looking at the costs of moving to hyperconvergence.
As you might expect the market for supplying hyperconverged systems is dominated by suppliers like Cisco, EMC, Oracle and HP, though a number of smaller players are also active. Analysts at Gartner expect hyperconvergence to be the biggest growing sector of the integrated systems market, expected to reach $5 billion by 2019 – representing almost a quarter of the sector.
"We are on the cusp of a third phase of integrated systems," said Gartner’s vice president Andrew Butler in May of 2016. "This evolution presents IT infrastructure and operations leaders with a framework to evolve their implementations and architectures."
Hyperconvergence is still a relatively young technology, but it offers many potential advantages. It will allow companies to reduce their reliance on different computing and storage platforms. It also offers greater resilience by eliminating points of failure, less need for ongoing systems management, and easier scaling. Because it uses commodity hardware it also means less risk of being tied to a particular vendor.
Businesses will need to adapt to this world, something which involves integrating management, reporting and security tools with those used for the rest of the company’s existing infrastructure. Whilst a completely HCIS system may be laudable future goal, in the short term systems have to be able to work alongside and communicate with existing setups.
"HCIS is not a destination, but an evolutionary journey," says Gartner’s Andrew Butler. "While we fully expect the use cases to embrace mission-critical applications in the future, current implementations could still pose constraints on rapid growth toward the end of the decade."