First, companies had to learn to migrate applications from on-premises to the cloud. That resulted in ugly surprises on the monthly bill, on-premises repatriations, and other fun stuff.

Now, even as many groan from cloud growing pains, they’re hearing that they’re behind the curve if they’re not multicloud, which can mean multiple public clouds, on-premises cloud, probably Kubernetes, perhaps serverless, and who-knows-what-else by next week. It’s enough to make them miss the slower, saner change cycles of on-premises legacy technology. Are there any staples in modern cloud information technology to moor themselves to as trends come and go? 

“I think the pragmatic reality for the Global 10,000 is: You’re going to be a multicloud company whether you want to or whether you don’t,” said Armon Dadgar (pictured), co-founder and chief technology officer of HashiCorp Inc.

With cloud providers all offering their own unique set of services, it’s unlikely that a company will want to commit to just one. Being able to pick the “best-of-breed” compute instances, object storage, machine-learning capabilities, or what-have-you, offers a competitive advantage.

It’s the bit about making all these clouds gel that is still so confusing to many. What separates  legitimate multicloud from a random IT environment that happens to have multiple clouds? Is multicloud even a noun? Or is it a verb describing the way separate clouds connect with each other, or share data, or punt apps around? 

Dadgar would argue that it’s the last. For the past couple of years, it has been popular to say cloud isn’t really a place but an operating model. That may be an even more apt way to describe multicloud, which has no standard infrastructure base. Companies want the freedom to run any workload on any cloud.

But the methodology of doing that is not exactly programmed into the muscle memory of IT staffers yet. There are a number of theories about what, ideally, multicloud is; they tend to be thin on details concerning the nuts and bolts of practical, day-to-day operation. These include the stable, repeatable steps in a multicloud workflow; those, according to Dadgar, are the ground pegs to which complex, mutable multicloud should be tethered.

Dadgar spoke with John Furrier, host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, during the ESCAPE/19 event in New York City. They discussed the role of workflow automation in multicloud developer operations (see the full interview with transcript here). (* Disclosure below.)

This week, theCUBE spotlights HashiCorp in our Startup of the Week feature.

Multicloud myth vs. reality

There are about four distinct definitions of multicloud in currency today, according to Dadgar. One is founded on the notion of data portability; data itself will exist in multiple clouds at the same time. Another — what Dadgar terms the multicloud “Shangri-La” — prioritizes workload migration; the dream is that one will be able to push a button and move a complete workload from cloud A to cloud B. This is not only hard to realize technically, it’s also true that the grass isn’t always greenest in the valley of Shangri-la.

“I think for most organizations, that’s very hard to architect for,” he said. “And I’m not sure it’s actually practical for most organizations, because it means that you can’t really use any of the cloud’s high-value services. It means that you have to really architect everything for data portability, everything for [workload] portability.”

Still another definition hinges on workflow portability. This approach is concerned with standardizing on workflow — basically, a sequence of steps needed to run an app to any cloud. Deployment, testing, continuous integration/continuous delivery and monitoring all follow the same course no matter where the app lands.

“If I’m the CIO, I don’t want to invest in four different workflows,” he said. “I want to train my team on one. I want to have a common way of delivering it.”

With the long-term direction of multicloud uncertain, companies can at least depend on the multicloud workflow. It nails down a critical part of the overall scheme and won’t expire in 15 minutes. And when mastered, it boosts the DevOps team’s efficiency, according to Dadgar.

And sometimes H: The new hows for multicloud

Founded in 2012, HashiCorp has built its business around workflow technology for DevOps teams working in multicloud. Through its “Freemium” model, HashiCorp provides both open-source tools and commercial tools. They leverage abstractions and automation, enabling teams to provision, secure, run and connect cloud-computing infrastructure. Its technology addresses the hows of multicloud infrastructure — these are crucial for the infrastructure itself — or the whats to harmonize into true multicloud, Dadgar explained.

Some companies might pair cloud-native architecture with a traditional IT practice. But, increasingly, they realize that this approach will leave a lot of their investment’s value untapped. They’re looking to modernize the processes as well through DevOps, self service, agile delivery and the like. “I think that sort of modernization of the process is just as important as the modernization of the architecture from on-premises to cloud,” Dadgar stated. 

HashiCorp recently announced the HashiCorp Consul Service for Microsoft Corp.’s Azure; it lowers the barrier to entry for organizations to leverage service discovery or service mesh across mixed environments while offloading the operational burden to HashiCorp.

It also unveiled Terraform Cloud last September. It allows teams to collaborate on and automate Terraform infrastructure as code workflows through a web application. 

Holding onto workflow as winds of change whip

Dadgar feels that HashiCorp’s focus on multicloud processes removes the company from Silicon Valley’s startup rat race. It’s not bringing out the latest, greatest version of a single-purpose tool while some recent grads in a basement are developing its successor.

“If we say, ‘I’m going to be the greatest shop at delivering Java,’ and then Docker shows up, that’s an existential threat to your business,” he said. 

The market in 2012, when HashiCorp was born, bore little remembrance to the one we see today. “You had one cloud; you didn’t have Docker; you didn’t have containers; you didn’t have Kubernetes; you didn’t have serverless; you didn’t have infrastructure as code,” he pointed out. “What hasn’t changed is core workflow.”

Cloud 2.0 isn’t just multicloud — it’s multi-everything, according to Dadgar. Companies have a much larger, more diverse set of technologies to bring under control. This is why we’re seeing more investment in management tooling and processes, he explained. The multicloud-management market was valued at $1,198.4 million in 2016; it will reach $6,816.5 million by 2023, growing at a CAGR of 28.4% from 2017 to 2023, according to Allied Market Research. 

The companies succeeding in the multicloud market tend to focus on workload and workflow processes, Dadgar pointed out. They’re approaching the workload as something that exists in the cloud as a managed service; Cockroach Labs’ distributed database-management systems for multicloud fall into this category, according to Dadgar.

“Or, you have the workflow vendors who have said, ‘Great, I’m going to give you a common multicloud, DevOps way of consuming that and deploying that workload out there,'” he said. “Those are sort of the two patterns that work.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the ESCAPE/19 event. (* Disclosure: TheCUBE is a paid media partner for ESCAPE/19. Neither Slower, the sponsor for theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE