Basic data centre infrastructure design has followed the same basic recipe almost forever. Compute, storage, network and more recently the hypervisor are treated as discrete components to be individually managed, albeit with the same ends in mind.
However, a wave of new applications and a need for greater levels of performance has driven a need for a different approach, one which delivers better response times, more flexibility and freedom to scale quickly to better meet changing business demands. As such, IT professionals are coming unstuck managing infrastructure the same way they did 20 years ago. In answer, new architectural approaches are emerging which collapse all critical infrastructure components into a single platform. Known as hyperconverged infrastructure or HCI, this has fast moved from emerging concept to go-to solution for many forward-thinking businesses.
What is hyperconverged infrastructure?
Unlike a traditional architecture, HCI leverages common datacentre hardware, typically x86 servers, as part of a single platform where storage, network, compute and hypervisor are combined.
Typically delivered as an appliance, most HCI solutions are geared around a scale-out architecture, where new appliances (sometimes referred to as nodes) are added as the infrastructure needs to grow. Most HCI solutions are pre-engineered to work with the public cloud so can interact with these services in a hybrid cloud setup or burst into AWS, Azure etc for additional capacity should on-premises resource get stressed.
A single piece of management software, overlaid across the entire cluster, is then used to manage every aspect to create a software-defined environment. This is designed to deliver cloud-like agility and performance with the security of on-premises infrastructure.
HCI differs slightly from a converged infrastructure where all three tiers are still discrete components but are managed from a single interface.
Most of the major vendors have got in on the HCI act, with solutions from Dell EMC, HPE, Nutanix and Pure Storage to name just a few.
Why does this matter?
There are several reasons why a business would look to implement a hyperconverged infrastructure, one of which is the cost-effectiveness of the solution.
Moving away from a traditional architecture by converging all of the elements into a single platform reduces the physical footprint of your infrastructure, allowing you to operate within a small physical space and reduce your overheads for both physical storage space and power demands.
There’s also less demand on IT personnel with no need to utilise specialist expertise for each area of your infrastructure. Everything can be managed by one person, or a small team of people, allowing you the freedom to allocate valuable IT resource to other projects.
Another significant driver for adopting a hyperconverged infrastructure is the level of flexibility it delivers which allows you to better respond to changing business demands. Scaling a traditional architecture is often complex and expensive, both of which impacts the speed and readiness of the environment and is often limited by the weakest infrastructure link in the chain. As a result, businesses will often over provision their infrastructure which adds additional expense and demand on resource.
With HCI however, scaling out is as simple as adding additional appliances, which can then be configured in just a couple of hours, with new workloads implemented in minutes. This allows you to better respond to fluctuating demands without a need for wasteful over provisioning.
You’ll also benefit from improved application performance as the converged infrastructure removes bottlenecks created by the distribution of compute and storage across different and potentially disparate architectures.
The future of HCI is disaggregated
Despite delivering many benefits, especially when scaling, one drawback of many HCI solutions is that there is no way to add additional storage or compute resource independently. As appliances consolidate everything, if you need more compute, you are often still left having to add more storage as well and vice versa. Although scaling the entire solution is relatively simple, for some environments this can become expensive.
This has seen the rise of what is known as disaggregated hyperconverged infrastructure, or dHCI. Also known as Hyperconverged infrastructure 2.0, dHCI decouples storage, network and compute at the node level so that these resources can be added independently of each other within one converged environment.
This allows you to provision additional storage, compute or network resource as needed without spending on a full node with unrequired resources included.
An example would be HPE’s Nimble Storage dHCI which houses HPE ProLiant servers and Nimble Storage arrays within the same chassis, creating a hyperconverged infrastructure with the option to scale both storage and compute independently.
learn more about hyperconverged infrastructure, discuss your own requirements or start your journey to a more agile and responsive approach to your data centre, get in touch with one of our expert team.