In just a few years, sovereignty has gone from being an obscure topic to an essential one. Hardware costs are rising. Ambitious AI projects often start in the public cloud. These are all issues that can quickly become headaches when requirements, preferences, or costs change. How do you stay flexible? We discuss this with Jean-Jacques Kroesbergen, Regional Business Lead for the Netherlands at Dell Technologies, and Luc Costers, Regional Sales Director for Benelux at Nutanix.
“A few years ago, there was a major shift toward the public cloud; that trend has completely reversed,” says Kroesbergen. “A great many customers are asking for on-premises solutions. We’ve practically become specialists in that area.” Think of fixed frameworks centered on protecting your most critical data, applications, and processes. “The crown jewels,” as Costers describes them. “It’s fun to start with AI in the public cloud. But over time, it’s those crown jewels, namely, your customers’ data, that become crucial. They must be protected in accordance with GDPR and European regulations, and you don’t want to put that data out in the open.”
Purchasing your own hardware, especially for ambitious AI projects, sounds appealing. However, the prices speak for themselves. But, as Kroesbergen emphasizes: “Ultimately, those hardware costs end up in the public clouds as well, because they’re all based on the same infrastructure.” Many customers who are at the forefront of AI are actually choosing to move these workloads out of the public cloud. According to Kroesbergen, this pays off: “For AI workloads involving long-term and intensive use, a dedicated sovereign or on-premises solution can reduce the Total Cost of Ownership (TCO) by up to 80 percent compared to continuous use of the public cloud, while giving you more control over data and governance.”
Repatriation
According to Kroesbergen, customers want above all to scale up quickly with solutions and remain “in control.” This means there is an emphasis on clear management, enabling effective decisions about the configuration of their infrastructure. “With, for example, Nutanix as a hypervisor, a market standard, you get that speed, IT integration, and proven partnerships.”
Costers summarizes customer needs in three words: simplicity, control, and security. This manifests, for example, in a drive to avoid “shadow computing,” that is, the unauthorized use of IT computing power within one’s own organization. However, this offers no guarantee that such computing power will actually remain internal: cyberattackers now also have AI ambitions, for example by hijacking someone else’s hardware. “That’s how we end up receiving some very strange invoices,” Costers notes.
At Nutanix, the response to customers is often a counter-question. Customers need to ask themselves where their data is located, which processes they manage themselves, and how easy and secure it is to isolate their own IT environment from the outside world. After all, the latter is just as relevant for legal sovereignty as it is for preventing day-to-day downtime. This raises the following question: which data and which processes can be repatriated?
One platform
Answering that question is one thing; implementing a sustainable solution that actually makes the data and processes manageable is quite another. “You have to make that flexibility simple,” says Kroesbergen. “That’s what we do for the customer in collaboration with Nutanix.” As a result, sovereignty can be a mix of public and private cloud. Costers agrees: “The more you modernize an environment and adopt the simplicity and flexibility of, for example, the Dell-Nutanix solution, the more control you gain over your data.”
That solution revolves around Dell Private Cloud (DPC) and Nutanix Cloud Infrastructure (NCI). Both companies emphasize a “two-tier” approach: NCI combined with storage from Dell Technologies. Costers: “This allows us to protect our customers’ past investments and integrate them with compute.” From a single pane of glass, any cloud environment, from private to multiple public clouds, can be managed, with dedicated storage where it’s necessary, cost-effective, and relevant.
In addition to Dell PowerFlex with the Nutanix Cloud Platform and AHV, which allow compute and software-defined storage to be decoupled, Dell PowerStore offers an end-to-end NVMe storage platform with per-drive scalability and Dell’s 4:1 Data Reduction Guarantee.
Eliminating silos
In this setup, organizations not only reduce their cloud and data silos but also the number of specialists required for each platform. Kroesbergen and Costers agree that having all those specialists for every individual infrastructure is neither feasible nor affordable. The Dell Automation Platform provides full support for the Nutanix stack in terms of deployment, management, and lifecycle management, ensuring that expertise is available in one place.
Costers sees Prism Central, Nutanix’s management layer, as the final piece of the puzzle. “It shows you your total capacity; exactly where it’s located is a secondary concern. You don’t have a separate monitor for Azure, a separate one for on-premises, and a separate one for the edge. It’s all consolidated into a single, centralized overview.” According to him, anyone who purchases components separately will run into problems. “If you buy a hypervisor from one vendor, go to a second partner for containers, and to a third for AI, you’ll find yourself in a jungle with no overview.”
AI without unlimited invoices
That third component, AI, is where Costers’ concept of “shadow computing” most frequently comes into play. Dell is introducing the AI Factory to address this. It consists of hardware, services, and an ecosystem of partners, one of which is Nutanix. “To me, an AI factory is something you feed data into, and from which a business outcome ultimately emerges,” says Kroesbergen. Nutanix adds two layers to this, Agent Gateway and Private Inference, which are primarily about control. “We show who managed which data or did something with it. And we ensure cost control, so that the number of tokens consumed is managed, monitored, and limited in advance. No unlimited bills.” With microsegmentation, Nutanix also places the AI environment in a sandbox, “so you don’t have an AI breakout that invades other areas,” as Costers describes it. Part of this is still in its early stages: the token budgets in Nutanix Enterprise AI 2.8 are currently a tech preview.
For those who need GPUs, the bill doesn’t have to come all at once. Dell works with consumption-based models where the customer pays for part of the hardware upfront and pays for the rest based on usage, at today’s prices. Costers wants to take it a step further. Together with Dell, he aims to select Dutch partners who offer shared GPU capacity, so that not every organization has to purchase its own hardware. After all, most organizations don’t need these chips constantly; they only pay for themselves with sustained use.
Conclusion
Sovereignty is unattainable without a change in working methods. Dell and Nutanix view the management layer as the key to success in this regard. An organization hopes to handle capacity and lifecycle management only once, not separately for each private or public cloud.
The balance between cost and control must also become simpler. The visibility provided by the partnership between Dell and Nutanix plays an important role in this.
At the same time, both companies emphasize that this flexibility should not be limited to a single partnership. In their view, vendors must prevent customers from being locked into a single provider in order to offer true freedom of choice. Dell therefore collaborates with a broad ecosystem of partners and supports various hypervisors and cloud platforms in addition to Nutanix.
Both experts also note that the partnership between Dell and Nutanix already offers customers many opportunities to reduce costs, increase control over data, and maintain operational flexibility.
Read also: Dell brings PowerStore to the Nutanix Cloud Platform