How do system architects effectively choose between standard relational database instances and horizontally scalable enterprise storage systems inside Google Cloud? Furthermore, one architecture excels at handling transactional applications with localized users, while the alternative solves global scaling limits without introducing data inconsistency or manual database splitting. Why does this choice fundamentally determine your application's capability to survive high-concurrency traffic over a massive geographical footprint?