I’ve read many stories that say IaaS is the future of computing. I disagree. All IaaS does is take the ghosts of networks past and shove them into the cloud. Granted, there are a few benefits in IaaS—mainly that it’s not all legacy infrastructure running in cloud VMs. But when I think of the future of computing, and more specifically the future of cloud computing, I see SaaS offerings like Office 365 as the last “aaS” standing.
In other words, I see IaaS and PaaS as interim technologies, not as the long-term future. They’re cloud-based halfway houses for the datacenters that IT has long focused on. But the move to SaaS means you won’t ultimately need to run anything like a datacenter, whether locally or in the cloud.
The overall cloud trend is to reduce hardware on-premises and push services to someone else’s infrastructure. But there remain many in IT who believe they need some infrastructure for their current environment, mainly for control of implementation, security, and data. IaaS, as well as PaaS, is where IT can exercise that control by shifting where the computing happens while still owning it.
In other words, IaaS is really about maintaining compatibility with the legacy, on-premises approach to IT. The cost of running IaaS in production (as opposed to the dev/test use of PaaS) makes it prohibitive for many organizations, especially because you still have all the other traditional expensess, from software licenses to management overhead. By contrast, SaaS absorbs all those needs—and their costs—in a cheaper package.
You might argue that the growth of IaaS such as Amazon Web Services, Google Cloud, and Microsoft Azure prove that IaaS is an integral part of the future technology stack. I believe we will continue to see growth in the use of cloud infrastructure. After all, many systems and services aren’t available in a neat SaaS package, so you move your legacy on-premises world onto cloud servers. Or maybe you have spiky business that has you loving the IaaS world because you can quickly provision for a holiday sale, then pull back when the spike ends.
Both use cases make sense now, but over time they will not. A SaaS offering like Office 365 will very likely be more sophisticated and feature-rich than what you develop yourself—so why not use SaaS wherever possible? SaaS can also handle the spiky demand benefit of IaaS. After all, what you need is on-demand scaling, which can be as easily provided as SaaS rather than your own IaaS effort.
For those of you moving to Microsoft’s cloud portfolio, the counterargument could be that Office 365 uses Azure Active Directory to work, so Microsoft is ensuring that IaaS is here for the long term. But that’s not IaaS—it’s really Office 365 Active Directory, which is SaaS. The IaaS label is more about semantic comfort than anything, helping IT equate the traditional on-premises Active Directory environment with an analogous cloud term.
Over time, more “infrastructure” services will become software services. Because once it’s SaaS, the boundaries between infrastructure, platform, and software don't matter to the IT customer—it’s merely a service. That’s a mental shift from IT’s on-premises view, where the boundaries matter in how IT delivers the ultimate service. Those boundaries will still exist for the provider, but won’t be IT’s concern.
That’s why I no longer argue over whether a client should choose Amazon’s, Google’s, or Microsoft’s IaaS. In five years, it won’t matter.
One day, we’ll drop all the infrastructure and plug in to the Amazon, Google, and/or Microsoft service fabric in whatever way we choose. Whether we call it SaaS or simply “service” won’t matter.