IT Equipment and Workload Management
What this domain tests
This domain covers how IT decisions - hardware selection, utilization, virtualization, and workload placement - translate into facility energy. The examiner wants you to connect server-level behavior to data center-level metrics.
Core concepts:
- Power proportionality: server power does not scale linearly with utilization. Idle servers still draw 40–60% of peak, so consolidating lightly loaded servers saves far more than the utilization numbers alone suggest.
- Virtualization and consolidation: raising average CPU utilization from, say, 10% to 40% typically allows removing several physical hosts, cutting both IT power and the cooling and UPS overhead that follows it.
- Right-sizing and decommissioning: zombie servers and over-specified hardware are common findings.
- Power capping and power management: enabling vendor power management or BIOS power profiles can reduce draw, but watch for performance trade-offs and latency-sensitive workloads.
- Workload placement: matching workloads to efficient hardware generations, and using scheduling or geographic load shifting to exploit cooler hours or renewable availability.
- Refresh economics: newer hardware often delivers more work per watt, so refresh cycles affect PUE-adjacent metrics like IT power density.
Traps to watch:
- Assuming a 50% utilization server uses 50% of its peak power.
- Forgetting that IT savings cascade: every watt saved at the server also removes cooling and distribution overhead.
- Ignoring performance impact when recommending aggressive power capping.
- Confusing server nameplate power with measured draw.
Worked example
Twenty servers each draw 200 W at low utilization. Consolidating onto eight servers at 300 W each gives 20 × 200 = 4,000 W before and 8 × 300 = 2,400 W after - a 1,600 W IT reduction. At a PUE of 1.5, facility savings are 1,600 × 1.5 = 2,400 W, which is the number to report.