How does the system determine the need for scaling and creating an additional cluster?

The system determines the need for scaling and creating an additional cluster based on monitoring key metrics and auto-scaling rules. Usually, the following parameters are considered:

• CPU and memory load on cluster nodes.
• Number of active requests or connections.
• Service response times.
• Task or message queues.

If metrics exceed predefined thresholds for a certain period, the system initiates scaling — adding new nodes or creating an additional cluster to distribute the load.

In Kubernetes, for example, the Horizontal Pod Autoscaler (HPA) scales the number of pods based on CPU or custom metrics. To scale the cluster (e.g., add new nodes), the Cluster Autoscaler is used, which reacts to resource shortages.

Thus, scaling is a response to monitoring and rules that define when current resources are insufficient to maintain performance.