Hyper-converged infrastructure (HCI) is a software-centric architecture that integrates storage, computing and virtualization resources in a single system that can run on off-the-shelf server hardware. It virtualizes the elements that are found in conventional systems, and usually includes a hypervisor for virtualized computing, a virtualized SAN (software-defined storage) and virtualized networking. It often provides an automated software-defined data center (SDDC) management system and a software layer that defines the operational aspects.
The difference between a hyper-converged versus a converged infrastructure is that in HCI both the storage area network and the storage abstractions are implemented virtually rather than physically in the hardware. This makes it possible to manage and configure all resources within the context of the hypervisor. This eliminates many inefficiencies found in traditional data centers and reduces the total cost of ownership (TCO).
The tight integration of components is what provides the key benefits of HCI: Ease of management and ease of scaling. When an organization wants to add more resources, it simply adds more nodes to the HCI. However, a drawback to this approach is that all resources must be increased in order to increase any single resource. This means that companies that want to get more storage capacity have to increase their compute power at the same time, whether they need more or not. Some newer HCI platforms are working to address that by offering nodes that are storage- or compute-centric.
HCI carries the risk of vendor lock-in, because you can’t combine nodes from one HCI vendor with those from another. Still, many organizations are drawn to the ease of management and the lower TCO that can be achieved in a hyper-converged data center.