We’ve been running 400-instance clusters of t3.medium on an on-demand basis during the work week for 5 weeks to complete our POC. For production, we’ll keep to the same schedule but I am concerned about getting those 400 instances up every Mon before 8am. My actual monthly utilization would be 65%. Given the size of the cluster, should I get a zonal RI with the capacity reservation? Will I end up paying more with the RI that assumes 100% utilization vs. going with on-demand rates with 65% utilization?