Yet Another "Why do cloud projects fail" article!

Some of my observations on the topic:-

Trying to make legacy applications work on cloud

As-is migration seems to be the path of least resistance, but can lead to expensive surprises, in certain cases, to a dead-end very quickly. Some of the common problems are bad code, lack of license portability, code with security vulnerabilities, no support for autoscaling, etc. Cloud is not a magic bullet to fix these problems.

Some of the familiar examples include:-

• Applications using softwares that do not have a cloud-friendly licensing model. Moving such applications to cloud might become cost-prohibitive.
• Application state stored locally on the app server
• Not leveraging caching later
• Code with security vulnerabilities

 Trying to make legacy process work for cloud

After you move to cloud, it does not make sense if you still require your teams to submit paper/online forms and go through 5 levels of email approvals before they can get a Virtual Machine provisioned and their application deployed.

Self-service provisioning, CI/CD, automation and integration with existing ITSM tools are not just concepts anymore. They are practical. Organizations moving to cloud without committing to improving the process (and make it agile) are only going to make the existing problem more complex to solve.

 Spending too much time discussing the "Art of Possible"

And not moving from Vision to Strategy to implementation quickly. Being bold is possible in cloud. But, many times the courage and commitment starts to fade away as we approach the implementation phase, organizations reverting back to the traditional thinking and approaches. 

How many times have you seen organizations/teams start with a high level solution using state-of-the-art, let's say, server-less architecture and later end up deploying applications on traditional VMs on cloud.

Sustaining the bold move all the way from Vision to implementation takes different mind-set and support from leadership.

Choosing bad use cases to prove the value of cloud adoption

Remember the t2.micros? With the entire application stack (including the DB) running on it, with a public IP assigned to it? This is what happens when an engineer without any clues about cloud, sets out to showcase the capabilities of the cloud, convinces his/her manager to setup an AWS account and creates a single instance of a small virtual machine (t2.micro?) and migrates one of the simplest web application (along with the DB) to that machine.

Choosing the first use case for cloud migration requires careful analysis.