pith. sign in

arxiv: 2605.16190 · v1 · pith:NGVAIFE5new · submitted 2026-05-15 · 📡 eess.SY · cs.SY· math.OC

Watts vs. Bytes: Turning Data Centers into Grid Assets via Storage Compute Co-Optimization

Shaohui Liu , Sungho Shin , Deepjyoti Deka This is my paper

Pith reviewed 2026-05-20 15:52 UTC · model grok-4.3

classification 📡 eess.SY cs.SYmath.OC
keywords data centersbattery energy storageco-optimizationgrid interconnection limitsworkload schedulingDVFSancillary servicespeak load management
0
0 comments X

The pith

Co-optimizing data center workloads with battery storage meets grid limits and doubles BESS value under stress.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a co-optimization framework for data centers with co-located battery energy storage systems under utility interconnection limits on peak load and ramping. It jointly manages deadline-constrained computing workloads through scheduling and dynamic voltage and frequency scaling together with degradation-aware BESS dispatch to reduce costs and enable ancillary service participation. Case studies using real-world market and workload data confirm that the framework produces feasible day-ahead schedules across operating conditions. Benefits increase substantially when interconnection constraints bind, with daily BESS value rising by a factor of two or more as storage actions reduce the risk of schedulable workload incompletion while respecting limits.

Core claim

The authors establish that a robust day-ahead co-optimization model integrating workload scheduling, DVFS, and BESS dispatch yields feasible schedules that optimize operations and grid services, with BESS value increasing by a factor of two or more under stressed peak-load and ramping limits primarily by mitigating potential incompletion in schedulable workloads while complying with constraints; under baseline conditions value stems from ancillary participation and improved management, while tight peak caps make workload composition matter such that non-schedulable jobs raise costs by more than 25 percent and DVFS emerges as an additional flexibility lever.

What carries the argument

The day-ahead co-optimization model that jointly optimizes deadline-constrained workload scheduling and DVFS with degradation-aware BESS dispatch subject to peak-load and ramping interconnection limits.

If this is right

  • Feasible day-ahead schedules are produced across a range of operating conditions.
  • BESS daily value increases by a factor of two or more when peak-load and ramping constraints bind.
  • Under tight peak-load caps a higher share of non-schedulable jobs raises operating cost by more than 25 percent relative to flexible mixes.
  • DVFS serves as a material flexibility lever when load limits are tight.
  • Coordinated compute-storage flexibility expands operational headroom and grid value of data centers.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Data centers could more readily locate in regions with constrained grid capacity if the approach is adopted.
  • The same co-optimization logic could extend to other flexible loads paired with storage such as EV charging hubs.
  • Real-time adjustments beyond the day-ahead horizon might yield further gains provided performance costs stay low.
  • Grid operators could treat data centers as more reliable providers of flexibility services.

Load-bearing premise

Workload scheduling and DVFS can be executed in real time with negligible unmodeled performance or reliability costs, and the chosen real-world market and workload traces represent conditions where interconnection limits actually bind.

What would settle it

Field implementation of the optimized day-ahead schedules in an operating data center that measures actual workload completion rates, energy costs, grid compliance, and any performance or reliability deviations from model predictions.

Figures

Figures reproduced from arXiv: 2605.16190 by Deepjyoti Deka, Shaohui Liu, Sungho Shin.

Figure 1
Figure 1. Figure 1: Overview of the robust day-ahead co-optimization framework for a data center with co-located BESS. The coordinator jointly schedules flexible compute [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Representative daily dispatch for a 100 MW data-center case with a 36 MWh/12 MW co-located BESS under hard interconnection constraints. The left panel shows the fixed load, internal data-center load with storage, and grid-facing net load with and without the co-located BESS. The right panel shows the robust SOC trajectory and scenario spread, together with charge/discharge and ancillary-service power compo… view at source ↗
Figure 3
Figure 3. Figure 3: Battery sizing results for the 100 MW data center with real PJM [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Optimal installed BESS energy capacity versus cycling limit for discount rates r=7% and r=10%. Megapack 3 remains at or near the search boundary, while Megapack 2XL is more sensitive to cycling and discount-rate assumptions because its higher power density is paired with higher energy￾specific capital cost. little beyond 1.0 cy/day. Megapack 2XL is more sensitive. At r=7%, its optimal installed capacity in… view at source ↗
Figure 5
Figure 5. Figure 5: Interconnection sensitivity under alternative load compositions over 31 daily instances. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Hourly LP shadow prices for the base interconnection case over 31 [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Battery value under stressed interconnection limits over 31 daily instances. The 36 MWh/12 MW BESS has modest value under loose interconnection [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Multiday discrete-DVFS case study under the 100 MW load-limit configuration. The cases compare no DVFS with 3-level ±5%, 5-level ±10%, 7-level ±15%, and 9-level ±20% operating sets. Overall, allowing DVFS improves the mean daily operating value and reduces day-to-day variability by providing additional scheduling and interconnection flexibility. by greater than 40%. In our case studies for stressed peak￾lo… view at source ↗
Figure 9
Figure 9. Figure 9: Multiday fixed-load and schedulable-demand ablation. In panel (a), scaling the fixed-load trace changes the residual headroom below the load cap; in [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Multiday out-of-sample robustness. The first-stage schedule from each 50-scenario training problem is evaluated on 500 fresh scenarios per day. [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
read the original abstract

Enabling continued data-center growth under increasing grid stress motivates closer coordination between flexible computing demand and co-located battery energy storage systems (BESS) to improve site operations and provide grid services. This paper develops a robust co-optimization framework for day-ahead operation of data centers with co-located BESS under utility-imposed interconnection limits on peak load and ramping. The model jointly considers deadline-constrained computing workloads, managed through workload scheduling and dynamic voltage and frequency scaling (DVFS), together with degradation-aware BESS dispatch to enable cost optimization and participation in ancillary-service markets. Case studies based on real-world market and workload data show that the proposed framework yields feasible day-ahead schedules across a range of operating conditions, with substantially larger benefits when interconnection constraints become binding. Under baseline conditions, BESS value is derived from both ancillary-service participation and improved workload and energy management. Under stressed peak-load and ramping limits, however, the daily value of BESS increases by a factor of two or more, driven primarily \revise{by BESS actions to reduce the potential incompletion in the schedulable workload while complying with interconnection constraints}. Under tight peak-load caps, workload composition also matters where a higher share of non-schedulable jobs can increase operating cost by more than 25\% relative to more flexible workload mixes. \revise{Additionally, DVFS studies further show that processor-level control is a material flexibility lever under tight load limits.} These results demonstrate that coordinated compute-storage flexibility can materially expand the operational headroom and grid value of data centers, especially under increasingly scarce grid capacity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript develops a robust co-optimization framework for day-ahead operation of data centers with co-located BESS under utility-imposed interconnection limits on peak load and ramping. It jointly optimizes deadline-constrained workloads through scheduling and DVFS together with degradation-aware BESS dispatch to minimize operating costs while enabling ancillary-service participation. Case studies using real-world market and workload data report feasible schedules, with BESS value rising by a factor of two or more under stressed peak-load and ramping conditions (driven by reduced schedulable-workload incompletion) and position DVFS as a material flexibility lever under tight limits.

Significance. If the central modeling assumptions hold, the work demonstrates that compute-storage co-optimization can materially expand data-center operational headroom and grid value precisely when interconnection capacity is scarcest. The use of real-world traces strengthens practical relevance and the finding that benefits amplify under binding constraints offers a concrete, falsifiable prediction for operators. These elements would be notable contributions to the literature on flexible demand and storage co-location if the quantitative claims survive additional validation.

major comments (2)
  1. [Abstract and Case Studies] Abstract and Case Studies section: The claim that daily BESS value increases by a factor of two or more under stressed peak-load and ramping limits rests on the assumption that jointly optimized workload scheduling and DVFS decisions can be executed in real time with negligible unmodeled performance, delay, or reliability penalties. When interconnection constraints bind, the model relies on these levers to create headroom; any real-time cost that increases incompletion or forces extra BESS cycling would directly erode the reported benefit and the assertion that processor-level control is a material flexibility lever.
  2. [Case Studies] Case Studies section: The quantitative results (feasible schedules, factor-of-two value increase, >25% cost impact from workload composition) are presented without error bars, sensitivity tables on key parameters such as DVFS cost coefficients or workload flexibility ratios, or explicit out-of-sample validation. This makes it difficult to assess whether the reported benefits are robust to the modeling choices that underpin the central claim.
minor comments (2)
  1. [Abstract] Abstract: The text contains visible LaTeX revision commands (e.g., “revise{by BESS actions…}” and “revise{Additionally, DVFS studies…}”) that should be removed before final submission.
  2. [Model formulation] Model formulation: Notation distinguishing schedulable versus non-schedulable workloads and the precise definition of interconnection ramping limits could be introduced earlier and used consistently to improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. The comments highlight important considerations regarding model assumptions and result robustness. We address each major comment below and outline the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and Case Studies] Abstract and Case Studies section: The claim that daily BESS value increases by a factor of two or more under stressed peak-load and ramping limits rests on the assumption that jointly optimized workload scheduling and DVFS decisions can be executed in real time with negligible unmodeled performance, delay, or reliability penalties. When interconnection constraints bind, the model relies on these levers to create headroom; any real-time cost that increases incompletion or forces extra BESS cycling would directly erode the reported benefit and the assertion that processor-level control is a material flexibility lever.

    Authors: We agree that the day-ahead co-optimization framework implicitly assumes that the computed workload schedules and DVFS decisions can be implemented in real time without significant unmodeled penalties or reliability impacts. The model focuses on generating feasible day-ahead schedules under interconnection constraints, with BESS actions explicitly reducing potential workload incompletion. Real-time execution dynamics and associated costs lie outside the current scope. In the revised manuscript we will add an explicit discussion of this modeling assumption and its implications in the Case Studies section, and we have updated the abstract to qualify the reported BESS value increase under the modeled conditions. revision: partial

  2. Referee: [Case Studies] Case Studies section: The quantitative results (feasible schedules, factor-of-two value increase, >25% cost impact from workload composition) are presented without error bars, sensitivity tables on key parameters such as DVFS cost coefficients or workload flexibility ratios, or explicit out-of-sample validation. This makes it difficult to assess whether the reported benefits are robust to the modeling choices that underpin the central claim.

    Authors: We concur that additional sensitivity analysis would improve assessment of robustness. In the revised Case Studies section we will include sensitivity tables for the DVFS cost coefficients and workload flexibility ratios. The presented results derive from deterministic optimization using real-world traces; we will add a short discussion noting the deterministic formulation and identifying stochastic extensions or out-of-sample testing as valuable future directions. These changes will better support the quantitative claims. revision: yes

Circularity Check

0 steps flagged

Forward optimization model with external data traces shows no circularity

full rationale

The paper develops a co-optimization framework for day-ahead data-center scheduling that jointly optimizes workload scheduling, DVFS, and BESS dispatch subject to interconnection limits. Case-study results are generated by applying this model to real-world market and workload traces; the reported benefits (including the factor-of-two BESS value increase under binding constraints) are simulation outputs rather than algebraic identities or parameters fitted inside the same equations. No self-definitional steps, fitted-input predictions, or load-bearing self-citations appear in the derivation chain. The framework remains self-contained against external benchmarks and does not reduce its central claims to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; standard assumptions of convex optimization and representative historical traces are implicit but not enumerated.

pith-pipeline@v0.9.0 · 5839 in / 1238 out tokens · 53516 ms · 2026-05-20T15:52:24.022099+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    2025 ERCOT system planning long-term hourly peak demand and energy forecast,

    ERCOT, “2025 ERCOT system planning long-term hourly peak demand and energy forecast,” Texas, USA, Apr. 2025, available at https : / / www . ercot . com / files / docs / 2025 / 04 / 08 / ERCOT-2025-Long-Term-Load-Forecast-Report.pdf

    work page 2025

  2. [2]

    Long-term reliability assessment,

    North American Electric Reliability Corporation (NERC), “Long-term reliability assessment,” NERC, Tech. Rep., 2023. 13

    work page 2023

  3. [3]

    Rethinking load growth: assessing the potential for integration of large flexible loads in us power systems,

    T. Norris, T. Profeta, D. Patino-Echeverri, and A. Cowie-Haskell, “Rethinking load growth: assessing the potential for integration of large flexible loads in us power systems,” Nicholas Institute for Energy, Environment & Sustainability, Tech. Rep., 2025

    work page 2025

  4. [4]

    Assessment of gaps in existing practices, requirements, and reliability standards for emerg- ing large loads,

    North American Electric Reliability Corporation, “Assessment of gaps in existing practices, requirements, and reliability standards for emerg- ing large loads,” Mar. 2026, nERC Large Loads Working Group White Paper. Available at https://www.nerc.com/globalassets/our-work/ guidelines/reliability/white-paper---assessment-of-gaps.pdf

    work page 2026

  5. [5]

    AESO connection requirements for transmission-connected data centres (2025 aug draft),

    A. A. E. S. Operator, “AESO connection requirements for transmission-connected data centres (2025 aug draft),” Calgary, Canada, 2025, available at https : / / aesoengage . aeso . ca / connection-requirements-for-transmission-connected-data-centres

    work page 2025

  6. [6]

    Using demand response to reduce data center power consumption,

    R. Hasegawa, “Using demand response to reduce data center power consumption,”Google Cloud Blog, October 2023. [On- line]. Available: https://cloud.google.com/blog/products/infrastructure/ using-demand-response-to-reduce-data-center-power-consumption

    work page 2023

  7. [7]

    Flexible data centers: A faster, more affordable path to power,

    C. Brancucci, D. Cutler, and J. Jenkins, “Flexible data centers: A faster, more affordable path to power,” Camus and encoord and Princeton ZERO Lab, Technical Report, Dec. 2025

    work page 2025

  8. [8]

    Data center power oversubscription with a medium voltage power plane and priority-aware capping,

    V . Sakalkar, V . Kontorinis, D. Landhuis, S. Li, D. D. Ronde, T. Blooming, A. Ramesh, J. Kennedy, C. Malone, J. Clidaras, and P. Ranganathan, “Data center power oversubscription with a medium voltage power plane and priority-aware capping,” inProceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and O...

    work page doi:10.1145/3373376.3378533 2020

  9. [9]

    Power provisioning for a warehouse-sized computer,

    X. Fan, W.-D. Weber, and L. A. Barroso, “Power provisioning for a warehouse-sized computer,”ACM SIGARCH computer architecture news, vol. 35, no. 2, pp. 13–23, 2007

    work page 2007

  10. [10]

    Managing cost, performance, and reliability tradeoffs for energy-aware server provisioning,

    B. Guenter, N. Jain, and C. Williams, “Managing cost, performance, and reliability tradeoffs for energy-aware server provisioning,” in2011 Proceedings IEEE INFOCOM. IEEE, 2011, pp. 1332–1340

    work page 2011

  11. [11]

    Expanding datacenter capacity with dvfs boosting: A safe and scalable deployment experience,

    L. Piga, I. Narayanan, A. Sundarrajan, M. Skach, Q. Deng, B. Maity, M. Chakkaravarthy, A. Huang, A. Dhanotia, and P. Malani, “Expanding datacenter capacity with dvfs boosting: A safe and scalable deployment experience,” inProceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1...

    work page 2024

  12. [12]

    Distributionally robust games for data center demand response coordination based on cpu utilization and quality of service,

    Y . Ding, T. Magnanti, and P. D. Martinez, “Distributionally robust games for data center demand response coordination based on cpu utilization and quality of service,” in2025 IEEE Power & Energy Society General Meeting (PESGM). IEEE, 2025, pp. 1–5

    work page 2025

  13. [13]

    Flexible operation of batteries in power system scheduling with renewable energy,

    N. Li, C. Uckun, E. M. Constantinescu, J. R. Birge, K. W. Hedman, and A. Botterud, “Flexible operation of batteries in power system scheduling with renewable energy,”IEEE Transactions on Sustainable Energy, vol. 7, no. 2, pp. 685–696, 2015

    work page 2015

  14. [14]

    Day-ahead operational scheduling with enhanced flexible ramping product: Design and analysis,

    M. Ghaljehei and M. Khorsand, “Day-ahead operational scheduling with enhanced flexible ramping product: Design and analysis,”IEEE transactions on power systems, vol. 37, no. 3, pp. 1842–1856, 2021

    work page 2021

  15. [15]

    An adjustable chance- constrained approach for flexible ramping capacity allocation,

    Z. Wang, C. Shen, F. Liu, J. Wang, and X. Wu, “An adjustable chance- constrained approach for flexible ramping capacity allocation,”IEEE Transactions on Sustainable Energy, vol. 9, no. 4, pp. 1798–1811, 2018

    work page 2018

  16. [16]

    Cost projections for utility-scale battery storage: 2025 update,

    W. Cole, V . Ramasamy, and M. Turan, “Cost projections for utility-scale battery storage: 2025 update,” National Renewable Energy Lab.(NREL), Golden, CO (United States), Tech. Rep., 2025. [Online]. Available: https://docs.nrel.gov/docs/fy25osti/93281.pdf

    work page 2025

  17. [17]

    The value of storage in electricity distribution: The role of markets,

    D. Lauinger, D. Deka, and S. Shin, “The value of storage in electricity distribution: The role of markets,”arXiv preprint arXiv:2510.12435, 2025

    work page arXiv 2025

  18. [18]

    Agent coordination via contextual regression (agentconcur) for data center flexibility,

    V . Dvorkin, “Agent coordination via contextual regression (agentconcur) for data center flexibility,”IEEE Transactions on Power Systems, vol. 40, no. 2, pp. 1832–1842, 2024

    work page 2024

  19. [19]

    Adapting datacenter capacity for greener datacenters and grid,

    L. Lin and A. A. Chien, “Adapting datacenter capacity for greener datacenters and grid,” inProceedings of the 14th ACM International Conference on Future Energy Systems, 2023, pp. 200–213

    work page 2023

  20. [20]

    Carbon- aware computing for datacenters,

    A. Radovanovi ´c, R. Koningstein, I. Schneider, B. Chen, A. Duarte, B. Roy, D. Xiao, M. Haridasan, P. Hung, N. Careet al., “Carbon- aware computing for datacenters,”IEEE Transactions on Power Systems, vol. 38, no. 2, pp. 1270–1280, 2022

    work page 2022

  21. [21]

    Flexibility from networks of data centers: A market clearing formula- tion with virtual links,

    W. Zhang, L. A. Roald, A. A. Chien, J. R. Birge, and V . M. Zavala, “Flexibility from networks of data centers: A market clearing formula- tion with virtual links,”Electric Power Systems Research, vol. 189, p. 106723, 2020

    work page 2020

  22. [22]

    Integration of flexibility from battery storage in the electricity market,

    J. Engels, “Integration of flexibility from battery storage in the electricity market,”IEEE Transactions on Smart Grid, vol. 11, no. 4, pp. 3270– 3279, 2020

    work page 2020

  23. [23]

    Coupling a small battery with a datacenter for frequency regulation,

    R. Guruprasad, P. Murali, D. Krishnaswamy, and S. Kalyanaraman, “Coupling a small battery with a datacenter for frequency regulation,” in2017 IEEE Power & Energy Society General Meeting. IEEE, 2017, pp. 1–5

    work page 2017

  24. [24]

    Opportunities and challenges for data center demand response,

    A. Wierman, Z. Liu, I. Liu, and H. Mohsenian-Rad, “Opportunities and challenges for data center demand response,” inInternational Green Computing Conference. IEEE, 2014, pp. 1–10

    work page 2014

  25. [25]

    Assessments of data centers for provision of frequency regulation,

    Y . Fu, X. Han, K. Baker, and W. Zuo, “Assessments of data centers for provision of frequency regulation,”Applied Energy, vol. 277, p. 115621, 2020

    work page 2020

  26. [26]

    Carbon-aware computing for data centers with probabilistic perfor- mance guarantees,

    S. Hall, F. Micheli, G. Belgioioso, A. Radovanovi ´c, and F. Dörfler, “Carbon-aware computing for data centers with probabilistic perfor- mance guarantees,”IEEE Transactions on Power Systems, 2025

    work page 2025

  27. [27]

    A unified framework for defining and measuring flexibility in power system,

    J. Zhao, T. Zheng, and E. Litvinov, “A unified framework for defining and measuring flexibility in power system,”IEEE Transactions on Power Systems, vol. 31, no. 1, pp. 339–347, 2016

    work page 2016

  28. [28]

    Modelling and characterisation of flexibility from distributed energy resources,

    S. Riaz and P. Mancarella, “Modelling and characterisation of flexibility from distributed energy resources,”IEEE Transactions on Power Sys- tems, vol. 37, no. 1, pp. 38–50, 2022

    work page 2022

  29. [29]

    Optimal bidding strategy for microgrids in joint energy and ancillary service markets considering flexible ramping products,

    J. Wang, H. Zhong, W. Tang, R. Rajagopal, Q. Xia, C. Kang, and Y . Wang, “Optimal bidding strategy for microgrids in joint energy and ancillary service markets considering flexible ramping products,” Applied Energy, vol. 205, pp. 294–303, 2017

    work page 2017

  30. [30]

    Chance constrained scheduling and pricing for multi-service battery energy storage,

    W. Zhong, K. Xie, Y . Liu, S. Xie, and L. Xie, “Chance constrained scheduling and pricing for multi-service battery energy storage,”IEEE Transactions on Smart Grid, vol. 12, no. 6, pp. 5030–5042, 2021

    work page 2021

  31. [31]

    Nantucket island energy storage system assessment,

    P. J. Balducci, J. E. Alam, T. E. McDermott, V . Fotedar, X. Ma, D. Wu, B. A. Bhatti, K. Mongird, B. P. Bhattarai, A. J. Crawfordet al., “Nantucket island energy storage system assessment,” Pacific Northwest National Laboratory (PNNL), Richland, W A (United States), Tech. Rep., 2019

    work page 2019

  32. [32]

    California ISO open access same-time information system (OASIS),

    CAISO, “California ISO open access same-time information system (OASIS),” https://oasis.caiso.com/mrioasis/logon.do, database, accessed 2025

    work page 2025

  33. [33]

    PJM data miner 2,

    PJM, “PJM data miner 2,” https://dataminer2.pjm.com/list, database, accessed 2025

    work page 2025

  34. [34]

    Flexible data centers and the grid: Lower costs, higher emissions?

    C. R. Knittel, J. R. L. Senga, and S. Wang, “Flexible data centers and the grid: Lower costs, higher emissions?” National Bureau of Economic Research, Tech. Rep., 2025

    work page 2025

  35. [35]

    Large-scale cluster management at Google with Borg,

    A. Verma, L. Pedrosa, M. Korupolu, D. Oppenheimer, E. Tune, and J. Wilkes, “Large-scale cluster management at Google with Borg,” in Proceedings of the Tenth European Conference on Computer Systems, ser. EuroSys ’15. ACM, 2015, pp. 1–17

    work page 2015

  36. [36]

    Resource central: Understanding and predicting work- loads for improved resource management in large cloud platforms,

    E. Cortez, A. Bonde, A. Muzio, M. Russinovich, M. Fontoura, and R. Bianchini, “Resource central: Understanding and predicting work- loads for improved resource management in large cloud platforms,” in Proceedings of the 26th Symposium on Operating Systems Principles, ser. SOSP ’17. ACM, 2017, pp. 153–167

    work page 2017

  37. [37]

    Analyzing Alibaba’s co-located datacenter workloads,

    Y . Cheng, A. Anwar, and X. Duan, “Analyzing Alibaba’s co-located datacenter workloads,” inProceedings of the 2018 IEEE International Conference on Big Data. IEEE, 2018, pp. 292–297. 14 TABLE I AGGREGATE SCHEDULABLE JOB PORTFOLIO USED IN THE REPRESENTATIVE100 MWCASE STUDY. WORKW j IS INMWH-EQUIVALENT COMPUTE-SERVICE UNITS,AND MAXIMUM RATEr max j IS INMW-E...

    work page 2018