Industry Flyover: Q4 2022
It’s the end of the quarter, the beginning of a new year, and we’re taking a look back at some of the news, opinions, and blogs we found interesting in Q4.
The Cloud Is Dead, Long Live the Cloud
It’s been a rough couple months for tech.
Inflation has soared, layoffs abound, and from coast to coast VC’s are tightening their belts and reshaping the startup landscape. The annual tradition of trimming budgetary fat has never felt so dire, and in these looming dark days even the cloud lacks its usual silver lining.
Basecamp has publicly abandoned the cloud. Startups are fleeing AWS due to cost. CIOs are wondering aloud “when will the cloud finally payoff?” while the industry scrambles to find an answer that will satisfy. Tech veterans have known for years that cloud costs are too high and too opaque for comfort, but eternal growth has staved off the reckoning we now face with varying degrees of horror and glee. So much for the cloud fairy tale. Yikes.
So, what is the problem with cloud costs?
Calling them “too high” and closing the book seems reductive. For years companies have been content publicly writing multi-billion dollar checks to cloud providers, and our own research on cloud usage indicated the problem with cloud billing had more to do with its opacity and unpredictability than any particular number. A big number on a cloud bill may be especially daunting in these trying economic times, but owning and operating servers isn’t exactly a bargain bin alternative.
The reason for the recent influx of cloud eulogy think pieces is multi-faceted, but if we were to describe it in one sentence or less it’d be: total cost of ownership. Or more accurately: a growing awareness of the total cost of ownership.
Cloud adoption through the 2010s was exponential and utterly transformative. From 2010 to 2020 the cloud market grew by 635%. That’s gold rush territory. It’s hard to capture the intensity of excitement surrounding the cloud in a few sentences, but the numbers speak for themselves. Cloud computing was such a significant step for tech as a whole that cloud adoption became a marker of organizational progress as much as actual technological progress. Fear of missing the innovation boat caused a spate of rapid lifts and shifts to public clouds, often with inadequate planning and a lack of financial oversight.
Under the threat of FOMO and with the vague promise of cost efficiency (a difficult promise to make even at the time) it was easy to hand-wave cloud costs as inevitable, or to attribute human operation costs to different sources, or to simply ignore the bills until someone from finance came knocking. This is especially true in ye olden days of Silicon Valley where velocity was king and no one asked questions so long as you grew fast enough.
What’s changed in recent months and years is not just shifting economic circumstances, but the more banal maturity of the technology. Cloud usage is commonplace. Best practices have emerged. Now there are thought leaders, industry reports, and tooling that didn’t exist fifteen years ago. There’s a level of awareness and scholarship that is slowly demystifying the cloud, and new disciplines like DevOps and platform engineering have shined a light on just how much work goes into supporting the cloud above and beyond the latest invoice.
Aside from the sticker price, there are dozens of services needed to stand up a cloud deployment. This sounds incredibly obvious, but there are an uncountable number of ways to deploy an API. Cloud costs aren’t unpredictable because of ignorance, they’re unpredictable because every application is different and when you go beyond a single region cost and complexity compounds. Eventually the complexity reaches a point where DevOps and platform teams are the only way to wire everything together and make it consumable to the wider team. When weighing the benefits of cloud computing against all the operational costs and human hours needed to make it work while scaling or at scale, it’s no wonder that some have declared the sky has already fallen. Historical data can be pretty unflattering with the benefit of hindsight.
But all this doomsaying, however warranted, is a symptom of a grander cycle of the cure becoming the poison becoming part of the cure (more on that later). Growing pains and public funerals are progress. A more critical stance on what an organization actually needs from its providers is ultimately a good thing for everyone and adopting a more strategic view of the cloud can and should be the order of the day in the new year. The cloud is no different than any other resource, and passing out of the honeymoon phase and into a utility phase is key to weathering this storm of cost concerns and calls for cuts.
The cloud isn’t magic, and holding your cloud spend to the same standards as, say, your marketing budget is just good business. Enabling that kind of strategic consumption, however, is easier said than done.
2023: The Year of Cloud Strategy
Admittedly “consume the cloud strategically” isn’t the most prescriptive takeaway from the section above, so what should we do with the cloud in 2023?
There’s not much we can say that David Linthicum hasn’t already said in his excellent “What's coming for cloud computing in 2023” article for InfoWorld:
“The problem most enterprises are having with cloud computing right now is related to too many cloud services that must be managed and tracked. Again, this leads to too much complexity, mostly through the rise of multicloud.”
Cloud complexity, all those aforementioned supporting services and people required to manage those services, is at the heart of ballooning cloud costs. It’s no great leap to say addressing that complexity, and therefore those costs, will be the focus of 2023 for the organizations who aren’t seeing the ROI they wanted but forgave in years past. The best way to address both cost and complexity is a matter for debate, but we agree with Linthicum that a layer above the cloud is the way forward.
"This cloud complexity problem can be solved through the strategic use of technology and better approaches to manage the complexity. Most important is reducing redundancy by using a common layer of technology above the public cloud providers as well as above any legacy or edge-based systems…We’re not attempting to solve every problem within the “walled garden” of each public cloud provider; this technology should exist within a common layer, aka supercloud or metacloud.”
We’ve actually written about superclouds before, specifically in the context of more strategic cloud consumption, and it reflects a broader shift in mindsets about what the cloud is and does. We believe the cloud is a utility, not an end point, and that all public providers are part of a seamless continuum that stretches out to the edge. Applications should use the optimal mix of resources based on user demand and location while developers use the tools they need regardless of provider. The cloud is not a goal, but it can certainly help you reach yours.
This isn’t a utopian ideal either. Making the cloud consumable and provider agnostic is the very basis for Internal Developer Platforms (IDPs) and comprises the driving force behind platform engineering as a discipline. It’s also core to our philosophy here at Seaplane. Be the change, etc.
It’s true that the cloud needs to change to serve the majority of users, but uncritical consumption was never going to be sustainable in the long term and it’s hardly “fair” to judge ROI based on hasty migrations. This moment of consideration, reevaluation, and even rejection is part of the maturing process. The hard truth is that some companies don’t need the cloud, or don’t need the cloud to change, and that’s fine. It doesn’t spell doom for the cloud as a whole, it just means that the industry is shifting into a healthier more thoughtful relationship with cloud resources.
The dream of the cloud may be dead, but the reality of the cloud is alive and well and changing for the better.
Edge and Superclouds
We can’t talk about the rise and fall and rise of the cloud without talking about the edge.
Edge computing plays a critical role in the cure-poison-cure-again cycle we touched on earlier, a cycle we predict will end with the proliferation of supercloud solutions managed with a layer above the clouds. We aren’t the first or last company to notice which way the wind is blowing, and this quarter we saw some big moves in the supercloud direction. Or "super platform" depending on who you ask.
Cloudflare published their thoughts about the edge as part of their Supercloud solution to cloud complexity, and how critical it is for supporting “Internet-scale” applications. Akamai and Macrometa announced their Edge Cloud Super Platform specifically to support global applications and simplify edge deployments. Hewlett Packard and CEO Antonio Neri’s “edge to cloud strategy” includes adding edge capabilities to their GreenLake hybrid cloud platform.
Even Michael Dell himself is a proponent of edge as an integral piece of tomorrow’s cloud.
It’s no coincidence that all the news stories this quarter covering the fall of the cloud coexist with stories about the potential of edge to be the hero we need in the new year. Calling edge “the next big thing” is behind the curve, edge is already the big thing moving to fill the gaps we now have the time and space to recognize within traditional cloud solutions.
Your Inflight Entertainment
You’ve heard of using your bitcoin mining rig as a space heater, but have you heard of using your space heater to mine bitcoin?
Well, you have now.
Happy New Year everyone.
That does it for this Industry Flyover. Thank you for flying with Seaplane, and we’ll see you next quarter!