“The common definition of multicloud is whatever the person talking is trying to sell you,” joked Duckbill chief economist and snarky-man-about-town Corey Quinn to former AWS executive Tim Bray on Quinn’s podcast. He’s not wrong, and it’s one reason that it’s sometimes hard to separate fact from fiction in multicloud. In fact, in that same podcast, the two (multi)cloud luminaries walked through how different vendors use multicloud to their marketing advantage.
Facts about multicloud
To start, it’s worth pointing out that multicloud is a simple reality for many enterprises. If you’re a company of even moderate size, you’re going to be running applications across multiple clouds. Sometimes this is strategy but sometimes it’s simply “enterprise IT.”
Also, if you work for an ISV like I do, multicloud is a business imperative, because customers want to run on different clouds. As Quinn highlighted, SaaS providers must run “in all of the major clouds because that is where their customers are.” He then went on to argue that game companies are also necessarily multicloud to manage latency: “A lot of the game servers themselves will be spread across a bunch of different [cloud] providers, just purely based on latency metrics around what is close to certain customer clusters.”
SEE: Research: Managing multicloud in the enterprise; benefits, barriers, and most popular cloud platforms (TechRepublic Premium)
So, yes, multicloud is a requirement for some companies, but maybe not yours. It’s also not especially easy to pull off, given the lack of expertise most people (and, by extension, their employers) have across clouds. Finding people who deeply understand more than one cloud isn’t easy, which is why Google’s Forrest Brazeal opined that it could be a real career accelerant to become proficient in more than one cloud.
Even so, many companies are going to try to build with a chosen cloud provider. Along the way, they’re going to hear all sorts of mixed messages from the different cloud providers.
Mapping multicloud to strategy
As Quinn pointed out, it is absolutely in the interest of every cloud provider except AWS to tout multicloud, because they know, Quinn suggested, that customers are not likely (yet) to be the “I’m all in on Cloud X” provider, “so they’re, of course, going to talk about multi-cloud.” But “if it’s AWS,” Quinn said, “where they are the 8,000-pound gorilla in the space,” the marketing sounds more like, “‘Oh, yeah, multicloud, terrible. Put everything on AWS. The end.’” AWS sometimes gets criticized for this, but “It seems that most people who talk about this have a very self-serving motivation that they can’t entirely escape,” Quinn concluded. He’s not wrong.
That said, things may be changing.
After all, a few years back AWS introduced ECS Anywhere and EKS Anywhere to enable customers to manage containers and Kubernetes wherever they wanted … including Azure and Google Cloud. It wasn’t as glaringly multicloud as, say, Google’s Anthos, but it still fit the multicloud description. And there’s been a softening of how AWS talks about multicloud, as Bray noted. After commenting that he’s “noticed that the [AWS] conversation online [about multicloud] has become much softer,” he reasoned this is because “going all-in on a single cloud is at least possible when you’re a startup, but if you’re a big company [like an] insurance company [or] a tire manufacturer … you’re going to be multicloud, for the same reason that [you] already have COBOL on the mainframe and Java on the old Sun boxes … .”
SEE: AWS Lambda, a serverless computing framework: A cheat sheet (free PDF) (TechRepublic)
In other words, because that’s just how enterprise IT works. No CIO in history has managed to exercise absolute control over the technologies running inside her company or the clouds her company keeps on retainer. AWS seems to be recognizing this, and while not embracing multicloud as a desired strategy, it has perhaps softened the message to reflect the reality of where its customers run their workloads.
Even if you do subscribe to the multicloud gospel, how do you get there? Paying someone to manage multicloud for you
As mentioned, it’s hard to find people who can straddle different cloud environments. For those who point to containers and, specifically, to Kubernetes to orchestrate their way to multicloud nirvana, this may be a case of the solution being as hard as the problem. As Bray argued, “[G]ood technology should make easy things easy and difficult things possible, and I think Kubernetes fails the first test there.” Why? Because “the complexity that it involves is out of balance with the benefits you get.”
Quinn piled on, saying he finds Kubernetes’ “complexity to be overwhelming.” So much so, he continued, that “finding someone who can competently run Kubernetes in production is a bit hard to do and they tend to be extremely expensive.” Similar to finding multicloud proficient people, in other words.
Which may be a good reason to let the cloud (by which I include IaaS, SaaS and PaaS) providers take care of all this multicloud complexity for you. Sure, you can figure it out. But it just might be cheaper and faster to pay someone else to figure it all out for you.
Disclosure: I work for MongoDB but the views expressed herein are mine.