Why the cloud computing hangover?

Costs, complexity, and lack of skills are prompting enterprises to press pause on their move to cloud.

recent survey by Virtana of 350 IT and cloud decision-makers found that 82% have incurred unnecessary cloud costs, 56% lack tools to manage their spending using automation, and 86% can’t easily get a holistic view of all their operational cloud costs.

Gartner predicts that 60% of infrastructure and operations leaders will see cost overruns for public cloud projects. This is assuming migration from traditional systems and cloud-native (net-new) development.

One could describe this as the cloud computing hangover. The move-to-cloud party is slowly coming to an end as we are forced to deal with the realities of this technology—the good and the bad.

The reasons for this hangover are obvious, and I’ve been covering them here for the past seven years. It comes down to efficiency, complexity, and reality.

Efficiency is getting the most value from your cloud resources for the least amount of money. This has been a problem for several years now. While cloud operations has been focused on keeping the newly migrated and cloud-native workloads and data up and running, it has not focused on managing cost efficiency.

This means that provisioned cloud services are not done away with at the end of their use, and the meter keeps running. Or, most common, resources are overprovisioned way more than needed to support a specific workload. Finally, companies are not taking advantage of cheaper options, such as reserved instances. 

The big one is not implementing cost governance systems that can resolve most of the problems just mentioned. Observability and automation services are purpose-built to optimize and reduce cloud costs. 

Complexity, as I’ve touched on many times, simply means that nothing is simple. Using many different cloud services for multicloud, and even single public cloud deployments, means that you’re way more heterogeneous than maybe you should be. That heterogeneity must be matched by tools and talent to operate each specific service, and that’s where cost overruns occur.

This is most easily resolved by building with common services, such as security and operations, and not allowing each cloud migration or cloud-native development project to use whatever cloud service they want. Of course, you need to do this without crushing innovation, finding a path that is best optimized and considering what will give you the bang you need for the buck you have.

Finally, and most misunderstood is the reality, on reality’s terms. I know that those marketing cloud computing for the past decade have played up cost savings as the core reason to move to cloud-based systems, but that should have had a large asterisk next to it.

The facts: Although there is potential for better cost efficiency, it takes a great deal of work to implement the culture, cost governance, skills augmentation, and other things that are needed to maximize cost optimization. Many enterprises have not been willing to do this. They have yet to grasp the reality that cloud computing requires a great deal of planning and ability to implement new skills and tools at the same time.

The issue now is that what’s been implemented “works,” and thus seems successful. However, you’ve not been successful given that the value of the cloud systems you’ve implemented is at a negative due to the unforeseen cloud costs emerging now.

I don’t mean to scare you away from cloud computing because you really have no choice. The innovation dollars have been chasing cloud-based platforms for the past several years, and if you’re looking to hold on to more traditional systems, you’ll likely suffer “death by a thousand cuts.” Support for those platforms will slowly be withdrawn during the next few years.

This all means that we need to get smarter about the proper use of cloud computing. Perhaps you could circle back to implement things that should have been implemented, focus on changing the culture, and fix any issues that may have been caused by a lack of planning, including complexity mediation before and after.

This hangover was easy to predict. Let’s get to work on the solutions. It won’t be the “hair of the dog.”

Copyright © 2021 IDG Communications, Inc.

How to choose a low-code development platform