To maximize parallel upload throughput from your on‑prem datacenter to Cloud Storage, which resource should you increase?

Increasing your network bandwidth from your datacenter to GCP directly addresses the bottleneck for data transfer from your on-premises environment to Cloud Storage. More bandwidth allows more data to be transmitted concurrently, thus maximizing parallel upload throughput. Increasing the CPU size on your server might help with data processing or compression before upload, but it won't directly increase the speed at which data leaves your network. Increasing the size of the Google Persistent Disk on your server is irrelevant as it pertains to storage capacity, not transfer speed. Increasing network bandwidth from Compute Engine to Cloud Storage would be relevant if the data were already in GCP and moving between services, but not for initial ingress from an on-premises datacenter.