Question about generic model and schedule-estimator

[zhy76]

Please provide an in-depth description of the question you have: Karmada uses two types of resource evaluation methods during the scheduling process: 1. A generic model based on Cluster Summary and Resource Modeling, and 2. An accurate evaluation model using schedule-estimator. It appears that both methods require listing the nodes and pods of member clusters. I wonder the performance of these two methods in large-scale scenarios and if there is any difference. Since both methods list all nodes and pods, why not directly use the estimator method to obtain a more accurate resource evaluation? What is the reason for the existence of the resourceSummary method? What are the considerations here? What do you think about this question?:

Environment:

• Karmada version:
• Kubernetes version:
• Others: