Multi-cloud Formation Template

The advent of Infrastructure as code has streamlined the first step of operations, orchestrating the infrastructure, to a large extent. Encapsulation of the whole environment where a highly available internet scale application can be turned on with directives in an observable text file has taken the complexity out of delivering value to the end customer, in a repeatable way.

The representation of these infrastructure templates is offered by the three cloud musketeers in different formats. AWS offers it as Cloud Formation(CFN) Template, Microsoft Azure does it as Azure Resource Manager(ARM) template while Google does it with cloud deployment manager. 

Planned or otherwise, the journey to the cloud has already started within an enterprise, and multi-cloud is closer to reality. Almost every enterprise, has deployed multiple applications tapping into different cloud services, choosing different vendors for their specific best-in-class capabilities. With the recent announcement of Anthos, Google has become the first cloud vendor to realize and embrace it. The promise of consumption of multiple clouds within an application brings the ultimate flexibility to the masses.

The new peril which was fast emerging as legacy complexity was morphing into cloud complexity, with applications utilizing services from multiple cloud vendors. This to a large extent can be circumvented by coming together of multiple cloud vendors for future efforts on Anthos.

Anatomy of a modern application

The below figure represents typical modern applications and the components that make it up.

The next logical phase for multi-cloud consumption would be to provide the tools for application scaffolding across different clouds.

The tools will have to provide not only the infrastructure requirements but the requirements for Latency, Scale-out and Cost as well.

My prediction is that "MultiCloud Formation Template" as a separate track will be part of the cloud conferences in the next couple of years. Hopefully, the bisection bandwidth that can support inter-cloud application between different cloud vendors will be close to reality by then.

As explained in this post by Jankiram Anthos is a calculated risk by Google. Will it pay-off in the long run? My take is that it's the combination of weakness (inability to penetrate the enterprise) and strengths (the world of commoditized Machine Learning) is what's pushing them to take this unconventional stance.

Nobody wins by playing safe and eliminating the downside risk. It can be a spectacular success or a spectacular failure, I for one will be cheering for this endeavor.