Optimizing Energy Efficiency at the WLCG PIC Tier-1: From Predictive Job Scheduling to Real-Time CPU Control

          @misc{ scivideos_oai:cds.cern.ch:3025620,
            doi = {},
            url = {https://videos.cern.ch/record/3025620},
            author = {},
            keywords = {},
            language = {en},
            title = {Optimizing Energy Efficiency at the WLCG PIC Tier-1: From Predictive Job Scheduling to Real-Time CPU Control},
            publisher = {},
            year = {2026},
            month = {may},
            note = {oai:cds.cern.ch:3025620 see, \url{https://scivideos.org/cern-cds/3025620}}
          }
          

Abstract

To mitigate the rising energy demands and CO₂ emissions of large-scale scientific computing, this study evaluated energy-aware resource management at the WLCG PIC Tier-1 site. We initially used machine learning to optimize node drainage by routing short jobs during peak energy prices. However, this complex scheduling approach proved impractical, requiring constant retraining for fluctuating workloads. Consequently, we identified real-time CPU frequency scaling as a superior alternative. Unlike complex scheduling, adjusting CPU frequency is straightforward, delivers immediate power reductions, and ensures controllable performance degradation. Node-level experiments validated this approach, quantifying the exact relationship between frequency, compute performance, and power draw. By estimating facility-wide savings based on these results, this research delivers a practical, scalable framework for automated, center-wide energy optimization at PIC and other WLCG sites.

00:00:00 Slide 1
00:00:28 Slide 2
00:01:23 Slide 3
00:02:00 Slide 4
00:02:27 Slide 5
00:03:31 Slide 6
00:05:51 Slide 7
00:07:05 Slide 8
00:07:48 Slide 9
00:08:07 Slide 10
00:08:45 Slide 11
00:09:21 Slide 12
00:10:40 Slide 13
00:11:42 Slide 14
00:13:07 Slide 15
00:14:10 Slide 16
00:15:16 Slide 17
00:16:48 Slide 18
00:18:03 Slide 19
00:19:29 Slide 20
00:20:40 Slide 21
00:22:23 Slide 22
00:23:06 Slide 23
00:23:40 Slide 24
00:23:53 Slide 25
00:24:46 Slide 26
00:25:04 Slide 27
00:25:32 Slide 28
00:26:52 Slide 29
00:29:27 Slide 30