IT architecture is a phrase capable of giving many an enterprise CIO and business lead alike the shivers. It’s not hard to understand why – modern enterprise-grade IT systems are both absolutely necessary and highly complex.
Despite the fact that no modern organisation can exist without IT, no one has yet come up with a way to build and run sophisticated systems at scale in a way that pleases everyone. On the business side, executives tend to want simplicity, reliability and, perhaps above all else, cost-effectiveness. On the technical side, IT leads want quality, performance, agility and room to grow.
When these competing priorities lock horns, what you get is an endless search for what has become something of an El Dorado for enterprise IT – the perfect system architecture. The frustrations and minor conflicts that arise from this search suggest there may be no such thing.
However, one approach to enterprise infrastructure that has been gaining a lot of support is hyperconvergence. Often abbreviated to HCI (for hyperconverged infrastructure), it is a concept that relates specifically to how data centres services are organised and run.
Whether operated directly or outsourced to a third party, pretty much every enterprise-level organisation relies on a data centre for its IT services these days. The centralisation of shared compute, storage and networking resources provides enormous efficiency savings compared to trying to run separate assets at every site. It ensures that systems are provisioned consistently, can communicate freely with each other, and provides the capacity to handle the enormous volumes of data generated by the modern enterprise.
Service-oriented data centres
However, running a data centre remains complex and, despite the obvious benefits they provide, not as efficient as some people would like them to be. One issue raised with the traditional way a data centre is organised is the separation of its main functions – processing, storage and networking. In a typical data centre, servers are the main unit providing the compute resource, with data storage and networking systems tagged on. The thought is, by running all three in their own silos, it is difficult to optimise workloads and therefore achieve maximum efficiency.
Enter HCI. The whole idea of ‘convergence’ is to bring compute, storage and network functions together in single virtualised, software-defined units. Once these functions are abstracted from the physical infrastructure and packaged together, they can be run from a single platform and resources shared across all three as required. This is sometimes referred to as the commodification of data centre infrastructure, where the emphasis is on the services being delivered rather than the capabilities of the systems.
As a service-oriented concept, HCI has an obvious appeal in a world now well accustomed to cloud technologies and as-a-service models of provision. HCI’s core value proposition is to say, let’s look beyond the conventional limitations of the equipment and systems – let’s pool all available resources together, and use what we have dynamically to service our compute, storage and networking needs as required.
HCI may well be that rare thing in IT architecture – an approach that can be seen to have strong appeal to both CIOs and business executives. On the one hand, the consolidation of all data centre workloads both simplifies management and generates significant efficiencies, including to costs. The fact that HCI services are software-based, use generic hardware and are available as single-vendor solutions means they have a low cost of entry and compare favourably to legacy data centre costs.
From the tech side’s perspective, perhaps the biggest advantage HCI offers is flexibility and scalability. In order to scale up resources in a traditional data centre, for example, you would need to add extra servers, extra storage capacity and additional networking capabilities – three times the complexity as well as three times the cost. Because every HCI unit is configured to deliver processing power, storage and networking capabilities in one, all you need to do is bolt on extra units. This makes the converged data centre model much more agile in response to shifting demands.
Simplification also suits IT administrators because it opens the door to using more automation. The straightforward scalability means you can automate on-demand capacity increases, for example, rather than having to manually configure three separate functions to make sure they all work together.