Forecasting the demise of IaaS

You may think it is bold of me to forecast the impending decline of Infrastructure As A Service (IaaS) before it is even mainstream, but I expect that the takeup and consumption of IaaS offerings will decrease gradually from 2025, down to almost nothing in 2030. So, IaaS is dead.

Firstly, to define the terms, IaaS is where a Cloud provider is responsible for providing virtualised compute, storage and network hosting, and presents a bare operating system for you to upload your application installation. The responsibilities for managing the operating system may vary by Cloud provider (see Azure diagram here)

So, why do I predict that this product will decline, when it appears to be that “IaaS is the new black” that goes with everything and is fashionable to everyone? There’s nothing wrong with the concept of service, the execution by the service providers, or the desirability of the service by customers – prices will drop (in the race to zero) further as competition increases, and capabilities will be increased.

It’s the service

Whilst assisting companies in their journey to the cloud, I am increasingly finding that the analysis of business needs and functions identifies that the dependence on specific operating system and application editions or versions is reducing. The needs are less likely to be “Application X on Windows 20xx R2” and more likely to be “business functionality Z to service customers within criteria Y”.

When it comes down to it – IT provides services to a business. The operating system is just a vehicle to do that. The application stack is just a toolset. The business needs to do a job, and IT exists to service the business needs.

The demise of the infrastructure team

Many people can see the decrease in the requirements for a dedicated infrastructure team within an IT department – caused by virtualisation and consolidation, by more reliable hardware with centralised and automated management. It only accelerates with the utilisation of IaaS services – less infrastructure to manage!

I’m uncomfortable with this possibility, because I come from nearly 20 years of infrastructure experience, but it’s the way the world of IT is going…

Who wants to patch servers, anyway?

The biggest drawback of a multi-purpose operating system like Windows is that it needs to be patched and updated. Toolsets are deployed and procedures are developed – to manage the underlying infrastructure that is required for the application stack to run. Large chunks of budget and whole teams of people are required to keep the lights on – even before the business gets any benefit of the application.

If an organisation can off-load the responsibility of patching and maintaining an operating system, wouldn’t that be a valid choice for a business’ bottom line? Moving the schedule, the responsibility and the risk of patching to specialists (the cloud providers) is the better option.

The rise of PaaS

In the same vein, would it also be beneficial for the business to also off-load the responsibility of maintaining the application stack, too? Of course it is – almost all the major projects that I have been involved in have been around the upgrade of an application version; coping with the deprecation of features, business customisations, leveraging new capabilities, etc.

Platform as a Service will grow, and as businesses learn that they can save internal operational expenditure by moving this to external operational expenditure – development of new and replacement applications will more often be in a PaaS service. This comes with a change in approach to understand the requirements of Cloud, but it is the direction that applications are going.

History has shown this to come true in business – it’s not the technology, it’s the change in practice and business attitudes that can kill off an old approach. Computers did not reduce the need for secretaries – just because a manager can now maintain their own appointment calendar and type their own letters / emails does not mean that the technology itself killed off the need for a secretary. Think of the decline in Mainframe around the turn of the century – the rise in client server implementations was not because mainframe was inherently bad or invalid, the transition was triggered by the Y2K bug, and the requirement for a business to consider alternatives .

IaaS is dead man walking

So, this is my prediction – IaaS will be popular for a while, and then applications will be re-developed for a PaaS platform that will meet the needs of the business – not just follow the IT function. Then, services that have been the early adopters of IaaS will move to PaaS and eventually SaaS.

Please share your views in the comments!