1. Create an ingress firewall rule with the following settings: "# Targets: all instances with tier #2 service account "# Source filter: all instances with tier #1 service account "# Protocols: allow TCP:8080 2. Create an ingress firewall rule with the following settings: "# Targets: all instances with tier #3 service account "# Source filter: all instances with tier #2 service account "# Protocols: allow TCP: 8080

Install the workload in a compute engine VM, start and stop the instance as needed, because as per the question the VM runs for 30 hours, process can be performed offline and should not be interrupted, if interrupted we need to restart the batch process again. Preemptible VMs are cheaper, but they will not be available beyond 24hrs, and if the process gets interrupted the preemptible VM will restart.

Reference: https://cloud.google.com/solutions/correlating-time-series-dataflow
https://cloud.google.com/blog/products/data-analytics/handling-duplicate-data-in-streaming-pipeline-using- pubsub-dataflow
https://cloud.google.com/bigtable/docs/schema-design-time-series

Coldline Storage is the perfect service to store audit logs from all the projects and is very cost-efficient as well. Coldline Storage is a very low-cost, highly durable storage service for storing infrequently accessed data.