How do you know nonfunctional requirements efficiently at initial stages in software engineering?

Non-functional requirements that affect virtually all projects fit into one of the following categories:

• Performance
• Availability
• Capacity
• Continuity
• Security

Performance is a customer requirement and is generally obvious. The other 4 have your operations department as their customer. You can figure out all the less obvious non-functional requirements before you start coding by sitting down with your operations folks.

Essentially anything that the operational people need to install and configure the software, monitor it, manage capacity, and fail over are non-functional requirements.

All the above non-functional requirements are very difficult to add to a project if they are not engineered into a solution from the initial design of the architecture.