A production Cloud Spanner instance runs in a single region and you need the fastest improvement in query performance while following best practices. What should you do operationally?

Cloud Spanner best practices recommend scaling out by adding nodes when high-priority CPU utilization consistently reaches 65% or higher. This threshold ensures that the instance has sufficient headroom to handle peak loads and maintain performance without over-provisioning. Adding nodes distributes the workload across more resources, directly improving query performance by increasing processing capacity. While optimizing heavy queries is a good long-term strategy, adding nodes is the fastest operational improvement for an already stressed instance. A 45% threshold is too low and would lead to premature scaling, incurring unnecessary costs.