Experiencing Performance Issues After Migrating APIs from ECS to EKS

My colleague and I have been transitioning our company's APIs from ECS to EKS, and while we've got the Kubernetes setup mostly in place, we've hit some major performance snags during our testing phase. Each month, we run a batch environment, so we decided to use that time to test out traffic shifting by sending a small fraction of requests to EKS while keeping ECS operational. At first, everything seemed to flow smoothly, but as the data load ramped up, EKS performance took a nosedive. Although Nginx and our APIs show low CPU and memory usage, requests are timing out, exceeding our default 5-second timeout. Interestingly, ECS handles the traffic just fine, and we're using the same container image across both platforms. Additionally, we have Istio in place and rely on ingress-nginx for our ingress controller, while the APIs pull data from MongoDB. We're trying to troubleshoot this and would love insights from anyone who's faced similar issues when moving from ECS to EKS, especially where Istio is involved. I'll make sure to update here with our progress as we dig deeper.