Dell: Private Cloud becomes "disaggregated" infrastructure

Contents

At its home trade fair, Dell introduced new features for the Dell Private Cloud, the Dell Distributed Private Cloud, and the Dell Automation Platform. They target disaggregated data center architectures where compute, storage, and network are no longer operated as a coupled HCI unit but can be set up and scaled separately.

The Dell Private Cloud supports multiple cloud operating systems on PowerEdge servers and PowerStore storage for this purpose. VMware Cloud Foundation 9.1, Microsoft Azure Local, and Nutanix AHV with Dell PowerStore are mentioned. This addresses environments where multiple hypervisors, private cloud platforms, or multiple cloud stacks are operated. Dell justifies the new approach with, among other things, multi-hypervisor environments, reduced vendor lock-in, the ability to switch between cloud operating systems, and independent scaling of computing power and storage.

New architecture targets AI workloads

The approach is the opposite of HCI, where compute, storage, and virtualization are linked in a tightly coupled stack. This simplifies procurement, setup, and operation. HCI is accordingly widespread in many private cloud environments. However, the downside is that the scaling and modernization of resources are also linked. This can become a problem for AI and data-intensive workloads, as GPU performance, storage capacity, file and object access, network bandwidth, power supply, and cooling rarely grow at the same rate.

In addition, many AI applications and agents are still in the pilot phase, proof of concept, or limited departmental testing. Once a use case is rolled out into production, user numbers, inference load, data access, availability requirements, and security specifications can change abruptly. This makes infrastructure planning more difficult than for classic enterprise applications with relatively stable load profiles.

Automation Platform for Operation and Orchestration

A disaggregated infrastructure creates more freedom, but also increases the effort for planning and operation. An environment of separately scalable servers, storage, networks, and cloud stacks is not an all-inclusive package. Dell tries to limit the additional operational effort through validated designs, central management, and automation. To this end, several extensions to the Automation Platform were introduced, such as the combination of AI agents with a dialog-oriented user interface.

The new platform can install cloud operating systems on PowerEdge and PowerStore and operate them via a common management model. Automation Studio complements these functions. This allows companies to create their own automation workflows across infrastructure and applications. Dell is thus addressing not only initial installation and provisioning but also recurring tasks during ongoing operation.

HCI under cost pressure

In addition to technical flexibility, cost advantages also speak for disaggregation. In an interview with iX, Dell Fellow Onur Celebioglu said: “HCI systems are particularly under cost pressure due to their architecture, as they require a lot of DRAM and drives. Externally scalable storage systems can therefore become more economically attractive in certain refresh scenarios.”

Another point is hypervisor binding. HCI is architecturally closely linked to the respective virtualization stack. Those who choose an ecosystem usually remain bound to it. Disaggregated architectures with external storage loosen this tight coupling. Dell sells this decoupling as “optionality,” meaning freedom of choice for different cloud and virtualization environments.

Distributed Private Cloud for Edge Locations

For edge locations, Dell is expanding its Distributed Private Cloud, formerly NativeEdge. The platform is getting two-node HA clusters, automatic failover, and integrated zero-trust functions. This targets manufacturing environments, logistics centers, or other decentralized locations where latency, data residency, or fault tolerance argue against pure cloud operation, but there is hardly any local IT expertise available. Here too, Dell follows the same logic: infrastructure should move closer to data and applications, yet be centrally managed and operated automatically.

Dell's new private cloud strategy is not a simple return to the classic three-tier architecture. Rather, the manufacturer is trying to combine the flexibility of separate compute, storage, and network resources with cloud stack options and automation. Whether the Automation Platform will sufficiently compensate for the additional operational effort remains to be seen in productive operation.

However, one thing is clear: those who want to run AI workloads locally must expect volatile load profiles, rapid architectural changes, and significantly more complex data paths. Dell is responding to this with more disaggregation, accepting that the private cloud will need to be planned, dimensioned, and orchestrated more extensively again.

(wpl)