Multi-Cloud is Finally Here!

First time this year, multi-cloud enterprises, as a customer segment of Sumo Logic, have grown faster than any other segment: 50% Y/Y. What took so long? In my conversations with enterprises over the last 5 years, there was only one strategy for public cloud and it was multi-cloud. But evidence of multi-cloud usage was sparse at best. Data from our Continuous Intelligence Report in previous years didn’t find much to support that the strategy for multi-cloud was being implemented.

After hundreds of conversations with enterprise architects on this topic, the strategy made sense to me. And just to clarify, these conversations were with Global 2000 enterprises going through digital transformation and as a result of that transformation, either planning a public cloud migration or already executing on it. Furthermore, these conversations were largely centered around IaaS and to some degree PaaS, not so much SaaS. Three core requirements underpinning the enterprise multi-cloud strategy came through:

1. Enterprises want to use best technology stack for their workload needs. Not all clouds are created equal and cloud service providers have distinct strengths. For example, GCP is perceived to have advantage in machine learning, Azure in support for traditional windows workloads, and AWS in having broad managed storage and database options.
2. Enterprises operate globally and require best in-region coverage for their workloads. Global 2000 operate across, well, the globe, and public cloud providers have varied coverage in regions. Data sovereignty rules place restrictions where regional workloads can run. Sometimes the coverage is slim enough that for resiliency you might need to have footprint across two or more to ensure business continuity.
3. Enterprises want to retain leverage over cloud service providers. Enterprise strategy today is to avoid vendor lock-in and to enable choice and retain pricing power in the hands of the enterprise vs. handing that power over to the cloud provider. Over the last 30 years, lessons related to vendor lock-in have been learned and in this transformation enterprises are applying those lessons carefully.

Until this year, the above was mainly just the articulation of enterprise cloud strategy. And, for many enterprises, the first logical step was to figure out how to do one cloud first and thus most were running workloads in one cloud, while keeping the broader strategy in mind. This year though, something changed and put that strategy into overdrive. We wanted to see if we could find any more data that either explains or enriches the trend.

Long-tail of cloud service adoption

AWS has some 165 services but fewer than 10% are adopted by more than half of its customers (figure 1). Most of those services, like S3, EC2, RDS, ELB fall into general purpose compute, database, storage, and network. Many of the services in the long-tail are management & tooling or niche platform services, etc. This seems to indicate that enterprises are largely sticking to consumption of services that are available within all cloud providers and are avoiding adoption of services that might lock them into one cloud provider.

Kubernetes

Kubernetes is the technology enterprises are betting on to build and run portable workloads. We have been monitoring adoption of containers, orchestration and serverless technologies for 4 years now and until last year it was unclear which orchestration technology will dominate. Last year, it started becoming clear that Kubernetes is becoming a clear favorite. With that, we wanted to see if there is any correlation with its growth and adoption across enterprises choosing multi-cloud. Data shows a strong correlation between enterprise adoption of multiple clouds in production and usage of kubernetes in production (figure 2).

In conclusion - the data shows that enterprise multi-cloud strategy is taking hold and is being implemented. Enterprises are running production workloads in multiple public clouds and they are carefully adopting cloud services to avoid vendor lock-in and ensure workload portability. It's clear that enterprises are not only betting on containers managed by Kubernetes for cloud migration, but also to ensure they can optimize their IT environments for better value and cost efficiencies. Furthermore, multi-cloud approach also requires best-of-breed independent tooling for management and security of those multi-cloud environments and the new container and orchestration layers of modern applications.