Do you have a cloud operating model?

In my new role leading technology, one of the immediate challenges has been to ensure our cloud journey is sustainable. Thanks to a very progressive operations team led by Adrian, we are pretty advanced in our migration to the cloud. We have some 'on tin' applications, but most of our workload is either in a private or public cloud, and we have the ability to move workloads dynamically between our different cloud environments. Also, most of our analytics is built to 'burst' in the cloud, allowing rapid parallelisation of model development. We can now build models in hours when it used to take weeks.

From an operations and development perspective this is all very cool and is delivering real value to the business. From a financial perspective though, the challenge is to make it also cost effective. Is the cloud always cheaper than on premise?

The reality is applications need to be built for the cloud, from the ground up. This requires developers to work a lot closer with infrastructure engineers, and even with the build and release teams. Applications built 'pre-cloud' tend to be more monolithic and are inherently less cloud friendly with regards to data management. They often maintain some state (data on user sessions etc), and so are less able to be started and stopped on demand without losing customers sessions and causing poor customer service.

As a result, these applications can often be more expensive to run in a public cloud than not, a significant consideration.

To get the benefits of the cloud, we have built a Cloud Operating Model that clearly works through the benefits of on premise and in cloud operations for all our applications. The result is we are a lot clearer on what applications to move, where to move them (private or public cloud), and even what environments to move (development, test or production).

The cloud operating model helps us to think about development holistically, rather than in the silo’ed approach common in IT, with separate concerns for development, testing, and operations. This forces us to be more ‘devops’ aware. It also helps with legacy applications and how we refactor these to work in the cloud, and the business case for this refactoring.

I believe good IT management is about optimising: optimising Capex vs Opex, optimising run and maintain vs innovation, and optimising build new vs develop existing vs buy. Our Cloud Operating Model is also an optimisation question and helps to provide clarity across all of our IT thinking.