Today Amazon Web Services (AWS) dominates the cloud, in large part because it offers significantly more services at competitive prices. The problem with cloud pricing, however, for AWS, Microsoft Azure, Google Cloud, and others, is that the very flexibility that makes them appealing may ultimately be their undoing. As Server Density CEO David Mytton wrote, "Right now all the data is there for you to analyze but you have to do most of the work yourself" to cut costs.
If this doesn't feel like a problem, perhaps you haven't spent enough time trying to decipher your cloud pricing options. As 451 Research analyst Owen Rogers has stated, "[G]etting to grips with the origins of the universe is far easier than understanding which AWS reserved instance option is the best one."
This doesn't sound like it rings the convenience bell that has traditionally driven cloud computing uptake, making price predictability the next killer feature for cloud computing. In turn, perhaps what's needed isn't someone to match or beat AWS prices, but rather someone to make it easier to run workloads in the cloud without needing a PhD to control costs.
It's not getting (much) better
In 2013 Brandon Butler wrote, "Cloud pricing models are complicated, which makes purchasing decisions for consumers difficult and comparing across providers a challenge as well." Unfortunately, that same statement could be written today with minimal variation. The basic principle of cloud pricing—pay for what you use—remains the same, but various ways to save money (like AWS' Reserved Instances) actually foster complexity. Don't believe me? Try reading through RightScale's attempt to compare cloud pricing models across clouds: Even as instance types become more similar, rationalizing prices becomes harder.
SEE: Cloud computing policy template (Tech Pro Research)
No, it's not that pricing is opaque. It's not. Well, not in the traditional sense of "if you want to know the price, you're going to have to play golf with a sales guy" way. Instead, as Rackspace engineer Kevin Jackson said, we've confused cloud transparency with simplicity: "If you have to provide an application to provide pricing, one cannot argue it's simple. It's transparent, not simple."
As RightScale's analysis demonstrates, we can now figure out Committed Use Discounts (Google), Convertible Reserved Instances (AWS), per-second pricing (everyone, though it depends on whether we're talking about containerized instances (Azure), updated instances to Dv3, Ev3, and Fv2 families (Azure), and more. But, to Mytton's point, "Modeling [these different pricing scenarios] at scale...would require some deep financial engineering experience."
Which, by the way, you probably don't have.
Convenience and the cloud
The cloud wasn't supposed to be like this. Back in 2012, speaking about the rising power of cloud computing, Redmonk analyst Stephen O'Grady pointed out, "[F]or developers convenience trumps most other technology characteristics." Security? Take a backseat. Ongoing manageability? Join it. And so on. Convenience may not be 100% of the calculus, but it's a huge part.
As such, O'Grady also asserted that one of "the biggest challenges for vendors built around traditional procurement patterns is their tendency to undervalue convenience." Ironically, O'Grady's contention may increasingly be true of the cloud vendors that have undermined more traditional software vendors.
As Mytton noted, keeping even a remote semblance of a grip on cloud pricing requires constant supervision. "If you stop paying attention, an unexpected change in patterns of usage can blow through your budgets early on in the month." In his experience, companies regularly misuse services that increase their costs, not to mention "unusual traffic spikes" and under-provisioning of capacity to add to the escalations. Additionally, the high costs of inter-zone networking, the need to search for discounts, and other issues may make "Cost optimization...a full time job," he noted.
This becomes two full-time jobs as companies come to rely on different services running on different clouds. Managing the technology may be eased with something like OpenShift, but managing the ever-shifting nuances of different pricing scenarios is not.
Google, perhaps more than Azure or AWS, has taken the lead on simplifying costs with its automatic application of discounts for sustained use (an attempt to combat complexity surrounding AWS' spot, on-demand, and Reserved Instance pricing). As Google's Miles Ward said, the cloud "should just be cheaper and easier."
Increasingly, it's not, which may spell opportunity for cloud vendors hoping to tackle AWS' seemingly indomitable lead.
Again, Google has taken the early lead in cloud pricing simplicity. More is needed, however, and doing so could pay significant dividends. If we look at other industries, customers tend to be willing to pay a premium for convenience. In retail, Nordstrom offers "no questions asked" returns, and shoppers happily pay more for the privilege. In travel, Uber took away the need to futz with payment (or, formerly, a tip), and quickly bypassed the taxi (or rental car) industry.
There's reason to believe that enterprises, and the developers who build for them, will respond in similar manner. The more applications run in the cloud, the less developers (and IT that works with them) will want to have to go through the mental gymnastics necessary to figure out how much ongoing costs will be. We've lost our way a bit in the cloud, focusing more on new features than convenience. In 2018 I expect we'll see pricing convenience return as a differentiator, one that may aid Google and other contenders for the AWS cloud throne.
- Amazon's 'Death Star' will be fueled by fat enterprise margins for years, and no one is safe (TechRepublic)
- Stressed about serverless lock-in? Don't be (TechRepublic)
- These two vendors are most likely to bring Kubernetes containers to the enterprise (TechRepublic)
- AWS isn't the cheapskate's cloud, and Amazon doesn't care (TechRepublic)
- Oracle unveils "universal credits," license mobility for easy cloud consumption (ZDNet)
Matt is currently head of the developer ecosystem at Adobe. The views expressed are his own, not those of his employer.
Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.