Why Cloud Did Not Kill the Data Center

Migrating to a cloud model would reduce costs and let me focus on consumption pricing; reduce complexity by moving backend software and hardware support to the provider; and increase agility by letting my developers use all those nifty new tools that were emerging daily from cloud providers. The data center was heading the way of the dinosaurs!

All was good. Everyone was happy. Well, in theory anyway. Until reality kicked in.

Complication

Many people were adopting cloud services and every vendor claimed theirs was faster and easier. Even people without technical knowledge saw the cost advantages, leading to rapid cloud adoption. Here are just some of the numbers:¹

• Worldwide end-user spending on the public cloud exceeded US$ 494.7 billion in 2022, up from US$ 410.9 billion in 2021.
• Retail (96.9%), media & entertainment (94.9%), and finance & banking (92.8%) are the top industry sectors with the highest cloud use.
• 31% of organizations predict they will run more than 75% of their workloads in the cloud by the end of 2023, and 20% already do.
• 47% of organizations now say they consider the cloud as the first option when incorporating new technology or initiatives. Only 5% are moving back to on-premises.

The real driver though is the use of cloud providers for new technology and services. DevOps teams are almost exclusively cloud-based with a focus on containerized solutions, Kubernetes, and the like. The availability of new tools for these teams is almost always cloud-first, which drives further adoption. Enterprises trying to allocate development resources tend to focus on those projects with the greatest near-term user value or impact, and in most cases, these are cloud-based projects and services.

However, if you look at the words behind the words, a different story emerges. While retail and media exceed 95% in the highest cloud use, these are newer cloud-based applications that do not speak to the thousands of mission-critical business applications that haven’t been migrated to the cloud yet. Likewise, 31% of organizations run more than 75% of their workloads in the cloud. But if you ask how many of those workloads are considered production or business critical, the number drops considerably. And yes, cloud-first should be the marching cry for new developers, but not cloud only.

Result

But alas, while everyone was looking at new applications, new business benefits, and new customer experiences, little attention was paid to the cascade effect. Many of those workloads that were migrated quickly to the cloud did not migrate from dedicated hardware and networks, leaving a substantial amount of infrastructure behind.

After shifting workloads to the cloud, people realized that they still needed the original IT infrastructure to support the mission-critical operations left on-premises. This infrastructure included networks, switches, servers, software, storage, operating systems, and personnel in the data center. Although transitioning to the cloud provided some cost savings, this did not eliminate the need for active IT infrastructure, and it was not possible to shut down everything once it was fully depreciated.

But being the resourceful group that IT people are, we were able to repurpose older equipment and maintain growth in older on-premises production applications and new data warehouses. We also developed new critical applications that were not suitable for cloud deployment. And we found a new use for the floor space like adding GPU-intensive applications that required low latency, or supporting workloads with compliance requirements that demanded on-premises infrastructure.

This infrastructure supports a significant amount of production applications in traditional data centers. Applications that cannot be easily migrated due to the amount of code; the number of integration points between other applications; the latency requirements; audit and compliance issues; emerging regulatory controls; or security and data sovereignty issues.

Could these legacy systems be migrated as well? Yes, eventually. But someone must find the budget and build a strong business case (and risk assessment) to justify fixing what already provides business value. In today’s world of tight budgets and improved customer experience, the approved funding rarely falls on fixing the old, but rather on building the new.

So, the cloud is not a replacement for our data centers. It has evolved into yet another piece of our overall digital infrastructure that we (those poor geeks in the data center) need to be aware of and support. Like all good technologies, the key is not if you use cloud computing, but how you use it.

So, what’s next for data centers?

Although the on-premises data center may be a dinosaur on the path towards extinction, it’s important to note that the actual dinosaurs existed for 245 million years. This means there is still plenty of time to complete any necessary migration projects.

Data centers are still growing, but now they are spread across multiple sites including edge, IoT, geo-specific, and SBU-focused. This makes it challenging to track, monitor, and maintain these assets due to increasing complexity.

Enterprises have found value in using cloud services and have also found new value in their data centers. However, to manage everything efficiently, they must treat both as a single infrastructure instead of separate technology silos.

Gartner predicts that by 2027, 35% of data center infrastructure will be managed through a cloud-based control system, which is a significant increase from less than 10% in 2022. To prepare for this shift, I&O professionals should prioritize building cloud-native infrastructure within their data centers, migrating workloads from their on-premises facilities to colocation facilities or the edge, or adopting as-a-service models for their physical infrastructure.²

This trend underscores the importance of integrating different systems. At the very least, sharing information between different departments is necessary for effectively managing the constant changes that occur within a business.

While it’s important to track assets, it’s also crucial to track all their activity throughout their lifespan. This involves various tools such as ITSM, ITAM, ITOM, DCIM, network and cable management, CMDB, and workflow management. I don’t require frequent inventory updates anymore, but I need real-time updates using automatic asset discovery, along with predictive and prescriptive analytics. This requirement is pushing software vendors towards a SaaS model, which allows them to update products more quickly and with less disruption to their customers. It also offers more flexible pricing models.

Conclusion

Far from going the way of the dinosaurs, data centers are still an integral part of today’s digital infrastructure. However, they must evolve to meet new demands. To do this, enterprises need to treat their cloud services and data centers as a single unit instead of separate technology silos. This will help them manage assets more efficiently while also allowing for better integration between different departments. With predictive analytics tools and automated asset discovery, businesses can ensure that their infrastructure is always up to date with real-time updates. By leveraging these technologies now, companies can prepare themselves for the predicted shift toward cloud-based control systems in 2027.

References

[1] Luenendonk M. 30 Cloud Computing Statistics, Facts & Trends for 2023 [Internet]. 2022 [cited 2023 May 30]. Available from: https://www.founderjar.com/cloud-computing-statistics/

[2] Keen E. Gartner Says Four Trends Are Shaping the Future of Cloud, Data Center and Edge Infrastructure [Internet]. 2023 [cited 2023 May 30]. Available from: https://www.gartner.com/en/newsroom/press-releases/2023-05-16-gartner-says-4-trends-are-shaping-the-future-of-cloud-data-center-and-edge-infrastructure