EdgeCampus has 6 edge servers (16 cores each, 100 W idle, 250 W at full load). During the day, 40 services run with a total demand of 70 cores; at night the demand falls to 15 cores. Powering a server on takes 3 minutes; live migration of a service takes 10–30 s.
- Compute the power of (a) spreading the load evenly over all 6 servers and (b) consolidating onto the minimum number of servers with an 80 % utilization cap, for day and night. What is the daily energy saving (assume 12 h day, 12 h night)?
- Design a consolidation policy: when to migrate and power servers off/on (thresholds, hysteresis), how to choose services to migrate, and how to protect latency-sensitive services (CampusAR) from overload during transitions.
- Identify two risks and how to mitigate them.
- Propose an evaluation (simulation with traces): metrics, baselines and one hypothesis you would test.