A company uses a 4 Gbps Direct Connect dedicated connection (LAG) to connect to five VPCs in us-east-1. Each VPC uses its own private virtual interface (VIF) to the on-premises network. Users report slowness and the engineer notices spikes in throughput that saturate the Direct Connect for about an hour each business day. The company needs to identify which business unit causes the spike and resolve the saturation. Which approach meets these requirements?

The correct option addresses both identifying the source of the spike and resolving the saturation. CloudWatch metrics VirtualInterfaceBpsEgress and VirtualInterfaceBpsIngress are the correct metrics to monitor traffic per VIF, allowing the identification of the business unit causing the spike. Provisioning a new 10 Gbps Direct Connect and shifting traffic provides additional capacity without downtime, as upgrading an existing dedicated connection requires a maintenance window. Incorrect options: Upgrading the existing dedicated connection to 10 Gbps would resolve saturation but requires downtime, which is not ideal for a critical connection. Using ConnectionBpsIngress and ConnectionPpsEgress metrics would show total connection traffic, not per VIF, making it impossible to identify the specific business unit. A 5 Gbps hosted connection might not be sufficient for a 4 Gbps saturated link and is a different service type. Using ConnectionBpsIngress and ConnectionPpsEgress metrics again fails to identify the specific VIF causing the issue.