The first thing I'd consider for datastacks is data sources, how much it will grow, security needs, the transformations I need to do and the people who will be using the data. Choosing the right tools for the job can be easy, but the right tools for the right job are more the ones that are compatible and up to good maintenance. Consider the flow of data in the system, ownership of each step in the system, and monitoring of problems. Additionally, there needs to be an estimate of software and engineering costs. When considering architecture and determining what is really needed, an outside view, such as Datalance's, can be helpful.