PARTIES.py is the PARTIES controller described in the PARTIES paper. It takes config.txt as the input file, which includes the number of colocated applications, and name, IP address and RPS of each colocated application. Real-time tail latency of each application is exposed through an NGINX web interface, running on port 84 on the server machine.

The provided code includes management of only the compute-resources (cores, cache, and frequency), for easier understanding. Management of memory and IO requires more precaution; random allocation may result in applications being killed by the OS for OOM (out of memory), or hanging due to substantial I/O delay. The memory wheel is better removed for non-IO-intensive applications.

monitorN.py monitors real-time CPU and LLC utilization. It is not needed in the PARTIES controller since PARTIES doesn't rely on real-time resource usage. You can comment out related lines in PARTIES.py if no monitoring is wanted.