Virtualization Technology News and Information
Supermicro 2022 Predictions: Cloud Repatriation - Return of the On-Premise

vmblog predictions 2022 

Industry executives and experts share their predictions for 2022.  Read them in this 14th annual series exclusive.

Cloud Repatriation - Return of the On-Premise

By Michael McNerney, VP of Marketing and Security, Supermicro

Having taken the tech industry by storm in the 2010s, the adoption of the cloud has been almost universal. Every major industry now leverages some aspect of cloud architecture or takes advantage of cloud services - from scale-out, open source, software defined, dev-ops, and public cloud.

But the cloud is not a "one size fits all" proposition. Today, many businesses are starting to recognize that many elements of the cloud architecture aren't well suited to their business - and some of their cloud adoptions haven't worked out as planned. This is driving more and more companies to shift back toward on-premise infrastructure. This phenomenon is being called cloud repatriation.

As large enterprises and smaller companies are evaluating and optimizing their cloud strategies, there has been a growing trend in recent years for cloud repatriation. These businesses have determined that they should operate their own private cloud versus outsourcing workloads to a public cloud provider. And the drivers behind this decision for twofold: cost and innovation.

Public Cloud: The Phantom Costs

While the convenience of an instantaneous infrastructure that can scale on-demand per your business' needs is compelling, it comes at a cost - and can quickly add up.

There have been some well publicized stories about businesses being hit with surprises in their cloud bills. Besides the costs of the CPU usage, there are additional charges for storage, network traffic within the cloud provider, availability zones, and ingress and egress charges for data to and from your data center and to customers. The high costs of these services are turning away many businesses, despite the conveniences offered.

Today's more technically savvy companies are starting to understand their scale requirements over time and can better estimate their IT demands and buy versus build tradeoffs. In fact, once the CAPEX and OPEX are accounted for, the costs for establishing an on-premise (or co-located) data center have noticeably more predictable costs.

An advantage of using systems entirely under an organization's control is that the systems can be allocated for different departments, workloads, and performance profiles as needed. For instance, if a cluster of servers is used during the workday for analytics and delivering low latency responses to the workforce over the nighttime hours, these systems can be used for mechanical simulations, predictive analytics, or other non-latency sensitive tasks.

Larger scale enterprises can also quickly achieve economies of scale for their private cloud infrastructure. They can manage unexpected aggregate demand and leverage a multi-cloud architecture for demand spikes while running their core compute requirements on the best cost and performance optimized infrastructure. For smaller businesses, however, one would expect that they would be unable to achieve the necessary scale, making it prohibitively expensive for them to consider managing on-premise infrastructure. However, we have seen examples where even smaller company's on-prem infrastructure is cost positive.

We are starting to see more stories of enterprises repatriating workloads to save money. In addition, many businesses today have "crunched the numbers" and found building their own private cloud to be a better option. So how much can you save?

As with any TCO discussion, it depends on the specific customer and environment. However, educated estimates for most environments and workloads would place on-premise private cloud costs around 30-50% less expensive than public cloud bills. If these estimates are even close, it will behoove everyone to look closely at their cloud spend and see if repatriation provides compelling value.

On-Premise: The Rise of Innovation

The second driver for repatriation is innovation. A primary consideration for these businesses on the cutting edge of AI, HPC, and more is the ability and timeframe to deploy the latest data center hardware as needed.

When new servers, storage, networking, or infrastructure components are available, getting these newer systems that can produce more work per watt for specific workloads - improving performance while driving down costs. Specific applications require the most performant CPUs, while other tasks may be able to run on previous generations of CPUs. A similar argument can be made for the memory in a server, the network attached to a server, and the storage needed.

No two enterprises are alike when it comes to processes and IT infrastructure for either hardware or software. Each business has its unique workflows, system utilization, storage capacity, and network utilization that determine its specific hardware needs. Likewise, by adding software to the party-perhaps with custom modifications-you end up with a custom-made data center environment tuned for what the company requires to perform its business.

By implementing a private cloud with an on-premise data center, specific hardware combinations can be acquired for the tasks that need to be accomplished. For example, servers can be chosen with the right-sized CPUs, memory, and local storage that minimize costs while delivering the agreed upon SLAs that employees or customers require. Additionally, network attached storage can be acquired, configured, operated, and maintained for the workloads that the enterprise performs.

Ignoring the current 2021 supply chain issues, companies operating their own infrastructure have much more control over deploying such hardware for their workloads. Vertical integration and optimization of IT components will enable innovation for enterprises while allowing a highly tuned infrastructure to meet the computing needs of the enterprise. Through in-house expertise and intelligent purchasing decisions matched to enterprise workloads, IT costs can be decreased while also reducing E-waste.

Evaluating Private vs. Public Cloud

Looking forward, organizations worldwide are starting to make more thorough analyses when evaluating their cloud strategy. These businesses are now identifying and leveraging the tremendous benefits of cloud computing while optimizing hardware choices specifically for their workloads and environment. The stakes are high, so the potential cost savings and innovation enabled are materially significant in today's fast-paced competitive environment.



Michael McNerney 

Michael McNerney is VP of Marketing and Network Security at Supermicro. Michael has over two decades of experience working in the enterprise hardware industry, with a proven track record of leading product strategy and software design. Prior to Supermicro, he also held leadership roles at Sun Microsystems and Hewlett-Packard.

Published Thursday, December 16, 2021 7:30 AM by David Marshall
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<December 2021>