On my last project, we built a configuration/collection tool for sensors installed in remote locations. For certain operations, our failure mode was "the user will have to go online so the app can resolve things, then can go back to the remote location and try again".
After folks explained that this could require a multi-day, multi-tens-of-thousands-of-dollars journey for the customer, we re-architected parts of the system to be tolerant of offline failures but still resilient and secure and saved the customer a potential round trip back to the field.
After folks explained that this could require a multi-day, multi-tens-of-thousands-of-dollars journey for the customer, we re-architected parts of the system to be tolerant of offline failures but still resilient and secure and saved the customer a potential round trip back to the field.