When people used to think about cloud infrastructure, they thought about VM’s and moving them from higher cost co-lo’s or onsite to a cloud provider. The goal was cost arbitrage by exploiting the economies of scale offered by the providers. Indeed, this remains the core AWS sales model – 'we can save you 10-30% just by moving workloads, don’t make it complicated and don’t disrupt the business'. The AWS market share shows the success of this approach.
However, my current clients are thinking about cloud infrastructure in a new way, with more sophisticated, ambitious, and business transformational goals.
Given how consistently these new goals are appearing, we believe the cloud infrastructure market is entering a second stage of maturity. In this stage, the Google Cloud Platform (GCP) is gaining market share quickly.
According to Gartner this past year, Google is in a solid 3rd place in the public cloud market. As Google Cloud continues to outspend Amazon and Microsoft, we will continue to see their tech leadership increase in the space.
So, what are these new goals and why is GCP winning?
Broadly speaking, the goals fall into two categories:
Faster scaling and innovation
Enabling new revenue streams and service offerings
Faster Scaling and Innovation
To achieve faster scaling and innovation, our clients are looking to containers, and specifically Kubernetes. This approach to managing infrastructure has 4 core benefits:
Automation - instead of patching and maintaining 100s of virtual machines, Kubernetes allows you to upgrade a single core image and automatically redeploy
Auto-scaling - running Kubernetes in a managed context like Google Kubernetes Service delivers on the cloud promise of infrastructure that expands and contracts to meet demand cycles
“Sublinear” scaling of Devops labor - the two preceding benefits are enabling technology teams to gain economies of scale in labor
Portability - each of the public cloud providers offer managed Kubernetes, which allows our clients to avoid lock-in
Kubernetes is the most popular cloud orchestration system out there. It was originally created internally by Google to solve cluster performance challenges in the core search business. I think of it as another level of abstraction – where VMs allowed us to run multiple ‘servers’ across hardware, Kubernetes allows us to run application servers across VMs. That abstraction enables a new management layer to automatically scale up and down and automate deployments. Google open sourced the project in 2014 and it became the de facto standard for containers almost instantly.
One client who manages Drupal websites for 300 customers saw a massive decline in costs when he re-architected the platform to run on Kubernetes. Instead of patching 300 virtual machines, he has a core image that can be modified and deployed to automatically upgrade all of his clients.
Scaling saves costs, but it also saves time - time to spend on projects like goal #2: Enabling new revenue streams and service offerings.Originally published on February 07, 2019