I read this blog post by Brad Feld, it discusses among other things the point at which the cost of the Public Cloud no longer makes sense, and the number mentioned is $200K / month.
Many of our prospects are engineering teams within Enterprises. These teams are in the process of setting up environments that can be used by their teams for development / testing purposes. These prospects typically ask us whether it is better to buy hardware and set up the environment on premise or if it’s better to use Amazon / Azure etc.
While there are always questions on whether it is secure to host in the Public Cloud, the topic of discussion quickly turns to the cost.
We typically ask the following questions to help decide:
If the workloads are spiky – as you would see with a load testing setup or with production workloads, then the preference is the Public Cloud. If the workloads are constant (like developer workspaces or a regression test setup etc…) then an on premise setup is better.
If it is required for the teams to access their workloads from everywhere – within and outside of the company networks, then it is preferable to do it on the Public Cloud.
If the teams already have hardware that can still be used for setting up a self-service solution or a private cloud and there is lab space available, then it is preferable to go with the on premise setup rather than a public cloud.According to the blog post, $200K / month is a good number to check on getting the workloads on premise. It may be interpreted as anything less than that is good for the Public Cloud. I think that this is not a good rule of thumb. Even in cases where the infrastructure expenditure is less than $200K / month, it is better to answer the questions listed above before taking the call.
What are the views of the readers?