I started my journey into Hyper-Converged Infrastructure (HCI) some three years ago. My employer at the time was investigating options for upgrading their datacentre and the words Hyper-Converged Infrastructure came up. I thought, what was that? I had no idea. So the journey began.
At the time we chose a product called DataCore SANsymphony and ran it as storage virtualization software. This basically meant that the SANsymphony software treated a big bunch of disks as a SAN on a Microsoft Windows Server. This gave us the flexibility we needed at the time but it wasn’t Hyper-Converged.
The compute was separate.
So HCI is at its most basic, compute, storage and networking converged together into a single platform. Both the SAN and storage abstractions are defined in software, rather than in hardware and run on off-the-shelf servers.
This dramatically reduces the cost of management and administration by presenting a single interface to the system, as well as lower-cost hardware. This, in turn, allows more scalability and functionality to be built into the platform as well.
Running out of storage? Add a new node. Running out of compute? Add a new node.
The fact that each node added to HCI increases both storage and compute greatly simplifies the upgrade path. For example, needing to retire old hardware and add new nodes and take the old ones out after data has been migrated to them. This has little to no impact on the end-users and reduces the amount of effort required to implement the changes dramatically.
Who does HCI?
Most vendors have HCI solutions, either built in-house or sourced from other companies. Here are a few:
Lenovo – Nutanix, Microsoft Azure Stack HCI, VMware VSAN
Huawei – FusionCube for Cloud, Nutanix
HPE – SimpliVity, Nimble dHCI
Dell EMC – VxFlex, VMware VSAN, Microsoft Azure Stack HCI
NetApp – NetApp HCI
Cisco – HyperFlex
Most of these solutions also offer hybrid cloud solutions with their HCI as well. This allows compute or software services to be shared with a Public Cloud Infrastructures (PCI) provider such as Microsoft Azure, Amazon Web Services or Google Cloud Platform. Some allow the seamless movement of Virtual Machines (VM) between a datacentre and PCI. This obviously has implications for High Availability and Disaster Recovery and each HCI should be assessed on all of its merits, not just cost alone.
So which HCI would I recommend? The one that fits your purposes the best. They all have different feature sets, all have different costing models and you need to weigh up what you require. At Diverse, we have chosen Lenovo and are currently implementing a Lenovo Nutanix HX cluster for our datacentre.