Editor’s note: The following is a guest post from Nathan Thomas, VP of multicloud at Oracle.
Nearly everyone who works in enterprise tech has some idea of what multicloud is, but it means a lot of different things to different people.
My years of experience building cloud-native, hybrid and multicloud solutions has given me an up-close view into what is working well — and what isn’t.
Multicloud shouldn’t be a squishy concept; its potential for the industry and customers is too vast. In this piece, I’d like to propose a new definition of what a multicloud service is and encourage the cloud industry and its customers to aspire to that definition.
But first, let me give some context.
Not long after cloud and SaaS offerings became commonplace, customers began talking about multicloud. Customers got value from a range of cloud providers and saw the potential to mix and match those solutions for workload and geographic suitability or as part of leveraging existing commercial relationships.
Even 15 years ago as this shift happened, many cloud services were actually well designed at the core to enable these use cases, with their focus on public APIs and user-configurable network security.
But most of the work to stitch everything together fell to the customer, with cloud service providers focusing on competition rather than cooperation.
Hyperscalers, which hosted many third-party cloud solutions along with their own native services, viewed direct integration with other hyperscalers as too risky. They had concerns over the potential that customers would move valuable workloads away after they had invested so much in winning those workloads.
Thus, hyperscalers offered powerful and diverse toolsets, but ones that worked best for solutions closely aligned with their own platforms, with multicloud left up to customers to figure out. Moreover, customers couldn’t rely on their stitched-together efforts to remain supported or fully functional long-term, should a hyperscaler retire any of the services they used.
Are the cloud walled gardens falling away?
Today, however, I have some good news. The cloistered ecosystem I described in my opening is becoming less viable in an era where customers demand choice and flexibility from their cloud providers, both in terms of integration and innovation.
Multicloud is entering a new phase, in which the cloud vendors are actively working on how to enable these solutions for customers. Now, we regularly see announcements around one cloud provider’s specialized cloud services running in another cloud provider’s infrastructure, or enabling cross-cloud network connectivity and data transfer.
It’s great progress and a big win for customers, but frankly, it is just the start of what’s possible.
Customers pursuing a truly multicloud approach still see challenges with technical integration, but perhaps more importantly, experience significant fragmentation across clouds in procurement, contracting, consumption and support.
These are real obstacles to adoption and cause friction some enterprises just won’t bother overcoming. They also mean there isn’t really a single economically efficient cloud computing market — just segmented cloud services functioning in isolation.
The way multicloud should be
Many traditionally on-premises vendors got accused of engaging in “cloudwashing” years back as momentum built around this new way of consuming services, hoping to seem relevant. To keep everyone honest now, we need a new and concrete definition of a multicloud service.
I propose that a true multicloud service provides an out-of-the-box solution to customers that spans multiple clouds, is fully supported in a multicloud configuration, is procurable in a single transaction and provides a unified operational and availability model as a cohesive service.
This may seem aspirational. But there are already a number of services that meet this definition, including Oracle, AWS, Google Cloud and Microsoft Azure. So far, we’re seeing those services drive increased, not lower consumption. They also don’t prompt customers to move existing workloads to competing clouds.
So, let’s define a new multicloud world, where customers aren’t locked in with development and procurement friction. Here’s what that new world can look like:
- As an industry, we need to focus less on data having gravity and more on data having business value.
- Hyperscalers will start to compete based on the quality of the services and support they provide, as well as how much value they can help customers unlock as customers define their enterprise AI pipelines.
- Customers will begin building apps with the confidence that these will work across clouds, at both a business and technical level.
Customers I speak with are clearly in support of this new model. In addition, the positive results I’ve had working with close partners on multicloud efforts are a good sign the industry is moving away from a defensive posture toward multicloud that is also more than lip service.
Real R&D and creative capital is being put into today’s true multicloud services from all involved. And who is in the middle gaining the benefits? Our shared customers.
The sooner we can make this definition of multicloud the default instead of the exception, the sooner we can unlock the true economic potential of cloud computing.
