Which metric is typically monitored for triggering autoscaling?

CPU utilization is a commonly monitored metric for triggering autoscaling because it directly reflects how much computational capacity is being utilized by applications in a cloud or distributed environment. When CPU usage approaches a predefined threshold, it indicates that the system may be under stress, resulting in slower response times or degraded performance. As a response, autoscaling mechanisms can automatically increase the number of instances or resources to handle the additional load, ensuring that applications continue to operate efficiently without impacting user experience.

On the other hand, while metrics like disk space utilization, network latency, and thread count can also be important in their respective contexts, they do not typically serve as primary indicators for autoscaling decisions. Disk space utilization pertains more to storage capacity issues rather than computational demand. Network latency reflects the speed of data transmission, which is crucial for performance but does not directly relate to the need for scalability in CPU resources. Thread count indicates concurrency within an application but is also more closely tied to resource management rather than triggering scaling actions. Hence, CPU utilization stands out as the key metric for autoscaling in cloud infrastructures.