What considerations would you take into account when designing for high availability in Cargill data pipelines?
