In the middle of September, we had another fun Build Day Live event. This time, we deployed the NetApp HCI platform and the NetApp Kubernetes on top of the HCI. Be sure to take a look at all of the videos from the Build Day Live with NetApp in Boulder as well as the event home page. We all enjoyed our time in Boulder, a beautiful central city with the Rocky Mountains rising up directly to the west of town. I have wanted to look at the process of deploying this HCI platform since I was convinced that HCI isn’t really about having the compute and storage cluster on the same servers. The NetApp HCI uses compute-only hypervisor hosts, and then shared storage using the scale-out Element storage OS from the SolidFire platform. Other HCI platforms use storage inside the hypervisor hosts, and so the storage cluster overlaps the compute cluster. With NetApp, there are two separate clusters, but the management is unified, it overlaps.
When we first learned about HCI, the apparent value was in eliminating the storage array with its separate network and management. After a couple of years of working with HCI, I realized that the real value was in simplified deployment and operations. At Tech Field Day 16 in Austin, Adam Carter convinced me that this value proposition is more critical to HCI than having storage inside the hypervisor hosts. In this Build Day Live, we deployed the NetApp HCI and saw that it is as easy as any HCI platform and easier than some. Part of the value of separate storage and compute clusters is the ability to scale each cluster independently. We showed expanding the storage cluster and then expanding the compute cluster separately. Both the compute and storage servers fit into the same 4-in-2U enclosure, including mixing node types in the same enclosure. We even mixed different models of nodes in a single cluster, both the storage and compute clusters can combine different models.
The final element of the Build Day Live was deploying the NetApp Kubernetes Service (NKS) on top of the NetApp HCI, a new feature in preview on the latest HCI version. I was impressed with the level of automation for deploying both the system Kubernetes cluster and new clusters for developers. Management of NKS is through a public cloud service, and NKS can deploy Kubernetes clusters onto a variety of public cloud and on-premises platforms. Having one tool to deploy and manage Kubernetes clusters across all of your platforms enables more application portability. I would like to see more of NetApp’s Data Fabric strategy, which enables data mobility to go with the Kubernetes application mobility.