Net app and the cloud conundrum

Ned Bellavance
6 min read


There are few technology changes that have been as disruptive as the public cloud. All aspects of the tech industry have been impacted in some way, but the hardware industry in particular has seen a major disruption. The advent of cloud hyperscalers has skewed the server, network, and storage markets. Suppliers can have an entire quarter made or ruined by the decision of a single hyperscaler to purchase new gear for a datacenter. By the same token, the cloud hyperscalers now have outsized influence over pricing, and can negotiate heavy discounts by purchasing at massive scale. Add in the fact that enterprise IT is shrinking, and traditional hardware companies are in a bit of a pickle.

Vendors like NetApp need to address the massive shifts in markets by developing fresh products and services that will appeal to enterprise customers in a post-cloud environment. Let’s look at some of the biggest shifts that have occurred due to the prevalence of public cloud.

  • Low friction consumption of services
  • Consumption based spending
  • Capacity on demand
  • Management of hardware and base systems

These shifts are all part of what public cloud offers to the consumer, and it is one reason that organizations have flocked to adopt it. While cost is often cited as a reason to move, the reality is that convenience, scale, and outsourced management are probably the most compelling features. How does a traditional hardware vendor become competitive in a landscape ruled by the new public cloud principles?

That is the cloud conundrum. NetApp has a possible solution.

NetApp Kubernetes Service

When you think of NetApp, you probably think of storage. That is what they have been known for, and in fact they are a leader for primary storage in Gartner’s Magic Quadrant for 2019.

NetApp Primary Storage Leader

You’d be forgiven for wondering why on earth NetApp would create its own Kubernetes service if it is primarily a storage company. But let’s remember that NetApp is trying to bring the principles of convenience, scale, and outsourced management to a hybrid cloud world. NetApp has been expanding its solutions into the major cloud providers, with first party support in Azure and GCP and third party support in AWS. To make these solutions work in the cloud, NetApp had to create a deployment and provisioning process that was automated and scalable. Based on this post by Jonsi Stefansson, the acquisition of Greenqloud brought with it a Service Delivery Engine (SDE) that enabled the rollout of Cloud Volumes. It appears that the SDE leverages Kubernetes to run cloud native applications, and so NetApp ended up inadvertently building a Kubernetes management platform for their own internal use.

While I’m not convinced of the long-term viability of NKS, I am certain there is a need for Kubernetes management software on the market. Rancher Labs has seen great success providing a platform that can manage multiple Kubernetes installations across a plethora of platforms. I’m not going to compare and contrast the two offerings - I’ll leave that as an exercise to you, but suffice to say there is a significant amount of overlap. NKS is able to provision and manage K8s clusters in the major cloud providers, with support coming for on-premises installations. It’s the on-premises installations that throws me for a moment, because while the cloud providers are highly standardized and well documented, on-premises deployments are heterogeneous, wildly differentiated, and a support nightmare. How is NetApp going to offer NKS on-premises? How is NetApp going to offer the same convenience, scale, and outsourced management in the private cloud sphere?

NetApp HCI

NetApp HCI is a great way to break up the heterogeneity of the on-premises datacenter, and make it possible for them to offer NKS on a standardized platform. By controlling the hardware and software layers of the solution, NetApp can offer a consistent, managed experience to end users. Imagine that you are already using NetApp’s Cloud Volumes in Azure and GCP, NKS to manage your K8s clusters, and Trident to manage your K8s storage. Adding NetApp HCI gives you that same deployment and management model on-premises. Except… it’s on-premises. Which means you have to work with your local VAR, buy the capacity you “think” you need, install the hardware in a rack, and get it configured. Then you have to manage the hardware going forward. That doesn’t sound very convenient or simplified for management. It’s a vastly different experience than using the public cloud, and it includes several points of friction that both infrastructure folks and developers find aggravating.

As Neil explains in the clip below. Enterprises are looking to move to the cloud for the lowered friction of getting started and the movement from CapEx to OpEx type spending.

NetApp’s goal with NetApp HCI and it’s Data Fabric approach is to abstract away the hardware on the private cloud side and simplify adoption by providing a managed offering. I couldn’t find a clip that talked about it, but Neil indicated that NetApp HCI will be made available as something approaching hardware-as-a-service, that would truly be an OpEx cost to the finance side. NetApp would be on the hook for managing the hardware, providing service, and adding capacity in what Neil called “packs”.

The idea of a managed, on-premises private cloud is not new. There have been several attempts from other vendors to sell a private cloud solution that matches the needs of an enterprise buyer. The cloud hyperscalers are also trying to get into the market with solutions like Azure Stack, AWS Outposts, and Google Anthos. I believe this validates NetApp’s approach of creating a product offering that enterprises can use in their datacenter. I think it also has an advantage in being hybrid solution not tied to a specific public cloud vendor, unlike Azure Stack and Outposts. In addition, NetApp already has a foothold in many enterprise datacenters, which gives them a leg up on something like Outposts, which is an AWS OEM hardware solution. There are a few considerations for execution.

  1. NetApp needs to engage with decision makers outside of the storage silo to build interest
  2. NetApp has to tread lightly with their existing VAR partners to strike a balance between convenience and the channel
  3. NetApp has to prove they have the services and support chops for the entire solution

None of these are simple tasks to execute, and I would be curious to see what NetApp’s strategy is to conquer each one. If nothing else, NetApp’s CFD6 presentation and new line of products and services has given me much food for thought.