Hybrid cloud? Private cloud? Public cloud? Multicloud? How to choose

My article “Don’t waste your time with a hybrid cloud” has gotten many people seeing red, and I can understand why. Many enterprise IT shops, and most enterprise technology companies, have pledged allegiance to the hybrid cloud model. 

The intellectual dishonesty of cloud-architecture purists

Their reasons are self-serving: Large enterprise software and hardware vendors have to believe in hybrid cloud computing, because they have the datacenter side of things to think about, and that where they make money. Thus, it’s bad for them to admit that public clouds will have a role, even if most existing workloads will reside in private clouds or remain on traditional datacenter platforms. 

Enterprises IT shops track with this argument for many of the same reasons. Their motivation is not to sell or hold on to a traditional approach to computing, but to maintain control—the agility and cost efficiencies of public cloud be dammed. It’s intellectually dishonest.

Of course it’s not just hybrid cloud proponents that are sliding into intellectual dishonesty. People who promote an all-public-cloud architecture or a multicloud architecture (using more than one public cloud) are just as bad. Worse, the private-cloud-only dinosaurs are still out there, and their push for only that architecture is downright scary. 

The fact is that not all workloads can move to public clouds, so a good deal of traditional systems will remain at the end of the “great cloud migration” that is underway now. (For what it’s worth, I believe about 20 to 30 percent of workloads won’t end up in the public cloud.) 

It really is true that you need to pick the rights tools for your job

There are no bad cloud approaches. Hybrid cloud, multicloud, private cloud, and public cloud all have a place depending on your requirements. My answer to “Which cloud architecture should I use?” is always that it depends entirely on the technology and business requirements of your enterprise, and not on that article you read. 

You need do your homework to understand the data, the processes, security, and governance—as well as performance, costs, and operations—before walking down the aisle with a cloud solution. In other words, you need to do all that boring stuff that enterprises often overlook and make mega-mistakes over that end up costing them millions to fix.

Still, it’s key to understand that not all cloud architectures are equal: Some are inherently better than others, and if you choose a less-optima cloud architecture, you need to realize you’re doing that and understand its implications—even if it is the correct approach for a particular need.

For most enterprises, private and hybrid clouds are poor choices. However, pairing traditional systems with public clouds—the pragmatic hybrid cloud—is a valid and cost-effective alternative to traditional hybrid-cloud architecture. And don’t lose sight of the reality that public clouds bring significant value, services, and security to the enterprise; in 2017, it’s time to admit this is simply the truth, no matter what fears you may have held in the past.

I’m taken back by the emotions around the cloud-architecture issue. I do understand that enterprise software and hardware vendors are spending collectively many millions on marketing messages, and even more money rebuilding their technology. It is scary and perhaps risky for enterprise IT goes against these well-funded and passive aggressive marketing monsters whose messages are tailored to appeal to senior business management, not just IT. 

However, enterprises need to snap out of the polarization around cloud architectures pretty quick. There is no single answer. Picking one—whether for fear, submissiveness, desire for control, management pressure, or marketing strength—is a career-ending mistake, for technologists and business managers alike. The costs are huge, and the concrete you sink yourself into is vast.

If you get it wrong, there will not be a reset button—trust me. You’ll be stuck in the wrong computing architecture.