Read just some of the terms casually thrown around by IT professionals – hyper-converged infrastructure, technology stacks, nodes, silos – and it becomes clear very quickly that it’s easy to get lost trying to follow the logic of how it works.
There is a logic to it, and that’s the key, particularly when it comes to the first of those terms: hyper-converged infrastructure.
“What it does is gives IT the ability to scale up using existing nodes, by adding additional drives and by being able to scale out that cluster by adding additional nodes on to it. It’s a highly scalable architecture, which converts infrastructure that was previously a little bit rigid,” said Ronan Carey, regional sales director for Dell Technologies Ireland.
Put more simply, it allows a business to buy in the technology it needs and then add more technology as it sees fit without negatively impacting on the systems it already has.
That way it can grow at a pace that suits it, and if it needs to diversify and change direction, it will have an IT system that can change direction with it.
“Normal converged infrastructure is when you bring the compute, network, storage and virtualisation components of your IT systems into one system or stack,” said Carey.
“Hyper-converged infrastructure then takes all that and then places those components in a ‘node-based architecture’. A node can be easily connected to another node and what this means is that the entire system can be extended very easily.”
This level of what’s termed ‘granular scalability’ means that companies don’t need to worry about future-proofing their IT systems when they buy them. They don’t have to purchase for future requirements and can instead just purchase what they need today without being penalised or stuck with an outdated system that won’t expand to meet new needs.
Then later, if they need to, they can add to the system as desired, scaling up or out as business needs dictate.
“It’s very easy to increase what it is that you do, and it’s agile enough to allow you to change direction – it’s extremely flexible. An important point is that this idea of scaling up or scaling out, they’re actually two different things, so being able to do either without damaging the existing infrastructure is very helpful,” said Carey.
Dell Technologies can facilitate this hyper-converged infrastructure, as well as adding on additional functionality that is more unusual.
“We offer top-of-rack switching, which means that you can manage the entire stack from one ‘pane of glass’, or in other words from one monitor. Rather than taking a siloed approach where everything is self contained and not connected to each other, and building those systems out, if you do it this way, it’s a lot easier to manage,” said Carey.
“One of our customers manages two reasonably big data centres that are hyper-converged across two different geographical locations – and they do that with just five people. So it gives you the ability to manage a lot more infrastructure with a lot fewer people and to be very agile about how you do it.”
Sounds good, right? If you have to manage a large quantity of IT resources, then anything that makes that less complicated is surely a no-brainer. Not necessarily, according to Carey. Dell Technologies doesn’t always recommend hyper-converged infrastructure to clients – and the reason is down to workloads.
“Whether you should adopt it or not entirely depends on the workloads you need to do, because not all workloads are suited to hyper-converged infrastructure. If you look at Dell Technologies, we have a big portfolio of speciality server and storage devices and people buy those based on the workloads they want to do,” he said.
“Enterprise resource planning workloads, for example, are generally run on dedicated infrastructure. If you have a very I/O intensive workload, in other words an application that reads and writes a large amount of data, you would probably lean towards a three-tier architecture, or maybe a converged architecture where you have it all in one stack, rather than hyperconverged, where it’s modular.”
In other words, with hyperconverged infrastructure users gain in scalability but lose in terms of raw technical horsepower.
“So the answer to the question of ‘is it for you?’ is: that it depends on the workload. When I talk to my team and our customers, it’s about designing systems for the workload as opposed to trying to sell a one-size-fits-all approach. That’s why you’ll see storage vendors still selling that kind of infrastructure – there’s a market for it and there will be for some time,” Carey said.
So who is it ideally suited to? According to Dell Technologies, anyone running a lot of virtual machines or using VMWare a lot.
“For VMWare workloads, it’s absolutely perfect. Let’s say you’re building out cloud-native applications, then it’s a really good architecture to do that from. We talk about a multi-cloud world, and your on-premises cloud should be a hyperconverged one and then you have the ability to burst into the public cloud if you need it,” said Carey.