As discussed with Olivier Vernin, we can decrease the size of pkg.jenkins.io VM:
It's currently a m4.4xlarge up 24/24 (0.80$ per hour, 16 vCPUs - Xeon Haswell at 2.4 Ghz , 64 Gb of memory, disk is an EBS with 2 Gbps of dedicated bandwidth, and network performances as "high").
- Memory is clearly overallocated
- CPU could be decreased: it was oversized to handle network contention when in "peaks": switching to a much more
The goal of this task is to switch this VM to a smaller size, while keeping a high quality network for the outbound traffic (bandwidth mainly) and a CPU with less cores (4-8) and eventually clocked at faster rate per core to better handles network peaks.