Cloud Getting Expensive? That’s By Design, But Don’t Blame the Clouds

Cloud adoption has flourished under COVID-19, thanks to the twin tailwinds of work-from-home and digital transformation. But escalating cost is emerging as a top problem as customers seek financial relief from public cloud platforms, which have become the biggest corporations in the world.

When going into discussions with your friendly neighborhood cloud provider to set terms for the next contract, don’t feel bad if you get the sense that you’re completely outgunned and overmatched by a superior foe. The cloud vendors are well aware of who has the upper hand, and it’s not you.

“The paradigm has really shifted,” says Aran Khanna, a former AWS engineer who is now the CEO and co-founder of Archera, a provider of cloud optimization solutions. “Instead of going to Dell and going to HP and making those reps give you the best price with best financing terms, [the cloud platforms] are basically saying, ‘Hey here’s all the data on how you could purchase things. Here’s all pricing. We don’t deviate from this.’ And it’s a 2 gigabyte file!”

Clearly, no human is going to be able to comprehend all the different parameters that go into setting prices for compute, networking, and storage resources in the cloud, Khanna says. What size instances should you buy, and for how long? Should you go serverless? What about using the cloud vendor’s managed services?

Want to compare one cloud’s offer with the offer from another cloud? Good luck with that. There are no standard SKUs, and you’ll need to do a lot of the translation yourself, Khanna says.

“There’s no apples to apples in the contracts, and they provide different flexibility over different dimensions,” he says. “One provides flexibility over compute instances and serverless. One provides flexibility over storage and serverless. So you can’t really compare the different commitment times, even between these vendors, which makes it very difficult for someone to reason about this in a spreadsheet. There are like 50, 60, 70 different dimensions, frankly, that you have to then go and work over. And maybe 50% of them are irrelevant to your use case, but you don’t know that if you’re just a finance person or just the engineer.”

So what’s the result of all this engineered complexity by the largest, most technologically sophisticated companies in the world? For most companies, they overprovision, because the downside of buying too little compute, storage, and networking would be detrimental to their operations. That’s a pernicious little twist on the paradigm shift that may have caught you by surprise.

The cost of cloud computing was the number one concern cited by respondents to a recent survey conducted earlier this year by O’Reilly Media. Of the 2,834 subscribers who took the survey, 30% cited cost as the most important concern, followed by modernizing applications using public cloud containers at 19%, and optimizing cloud performance at 13%.

“The common perception that cloud computing is inexpensive isn’t reality,” writes Mike Loukides, O’Reilly’s vice president of content strategy in the new report, titled “The Cloud in 2021: Adoption Continues” (the report can be accessed here).

“Being able to allocate a few servers or high-powered compute engines with a credit card is certainly an inexpensive way to start a project, but those savings evaporate at enterprise scale,” Loukides continues. “The cloud provider will reap the economies of scale. For a user, the cloud’s real advantages aren’t in cost savings but in flexibility, reliability, and scaling.”

IT decisionmakers who try to replicate the old way of provisioning compute resources as their operations move to the cloud are likely going to end up on the wrong side of the equation when the costs and benefits are tallied up at the end of the quarter, says Khanna, who helped implement Sagemaker at AWS and also worked on provisioning GPUs on the cloud while working with Nvidia.