Performance degradation during peak hours is typically a direct symptom of resource contention, insufficient CPU, memory, or compute capacity relative to the demand placed on the VMs at that time. The most direct and effective remedy is to scale capacity to meet that peak demand, either by provisioning additional VM instances to horizontally spread the load or by allocating more compute resources (CPU, memory) to the existing VMs vertically, both of which directly relieve the resource contention causing the performance issue during high-demand periods. This capacity management response is the standard first action for peak-hour performance problems tied to resource saturation. Changing storage tiering and cache configuration (B) can meaningfully improve storage-specific I/O performance, but it addresses only the storage dimension of performance and would not resolve a broader compute (CPU/memory) bottleneck if that is the actual constraint during peak hours, making it a narrower and less universally applicable fix than adjusting VM resources directly. Increasing the number of network ports (C) addresses network throughput or connectivity capacity specifically, which is not necessarily the bottleneck implied by general VM performance issues during peak load, and does not address compute resource contention. Reducing storage size (D) does not improve performance at all and could actually constrain the VM further, making it the opposite of a helpful action. Adding capacity is correct.

Reference topic: Managing the Data Protection Environment - Capacity and Performance Management for Peak Demand.