{"id":{"repo_id":"potsdam-thes","oai_identifier":"oai:kobv.de-opus4-uni-potsdam:67320"},"canonical_url":"https://search.dev.ndltd.org/etd/potsdam-thes/oai:kobv.de-opus4-uni-potsdam:67320","repository":{"repo_id":"potsdam-thes","name":"Universität Potsdam - Thes","base_url":"https://publishup.uni-potsdam.de/opus4-ubp/oai"},"display":{"title":"Squirrel","abstract":"Despite significant advances in hardware efficiency, data centers continue increasing their electricity consumption due to rising application demand. Consequently, data center carbon emissions grow. In addition to energy-saving mechanisms, increasing progress in carbon-aware software is necessary to reduce carbon emissions and mitigate climate change. Enhancing workload managers used in data centers with carbon-aware mechanisms counteracts their emissions. State-of-the-art solutions do not cover carbon-aware scheduling for the widely used workload manager Slurm. This thesis develops and evaluates three carbon-aware scheduling algorithms for high-performance computing clusters using the Slurm scheduler. To this end, we elaborate on whether we can reduce operational emissions using spatiotemporal workload shifting within a Slurm cluster in Germany and how the effects differ across energy zones with different characteristics of their electricity grid. We develop and evaluate Squirrel, a carbon-aware batch job scheduler prototype that considers the grid carbon intensity and servers’ thermal design power values. We also propose a forecasting algorithm for grid carbon intensity that can be used when there is no access to forecasting data. We measure the energy profiles of different TPCx-AI jobs and use their hourly energy consumptions to simulate workload scheduling for three different cluster configurations. Our simulations compare Squirrel with carbon-agnostic first-in-first-out scheduling in five different energy zones. The comparison is based on the hourly computation of schedules using data throughout 2023. Squirrel achieves average emission savings between 0.3% and 32% in our simulations, demonstrating the practical implications of our research.","abstract_html":"Despite significant advances in hardware efficiency, data centers continue increasing their electricity consumption due to rising application demand. Consequently, data center carbon emissions grow. In addition to energy-saving mechanisms, increasing progress in carbon-aware software is necessary to reduce carbon emissions and mitigate climate change. Enhancing workload managers used in data centers with carbon-aware mechanisms counteracts their emissions. State-of-the-art solutions do not cover carbon-aware scheduling for the widely used workload manager Slurm. This thesis develops and evaluates three carbon-aware scheduling algorithms for high-performance computing clusters using the Slurm scheduler. To this end, we elaborate on whether we can reduce operational emissions using spatiotemporal workload shifting within a Slurm cluster in Germany and how the effects differ across energy zones with different characteristics of their electricity grid. We develop and evaluate Squirrel, a carbon-aware batch job scheduler prototype that considers the grid carbon intensity and servers’ thermal design power values. We also propose a forecasting algorithm for grid carbon intensity that can be used when there is no access to forecasting data. We measure the energy profiles of different TPCx-AI jobs and use their hourly energy consumptions to simulate workload scheduling for three different cluster configurations. Our simulations compare Squirrel with carbon-agnostic first-in-first-out scheduling in five different energy zones. The comparison is based on the hourly computation of schedules using data throughout 2023. Squirrel achieves average emission savings between 0.3% and 32% in our simulations, demonstrating the practical implications of our research.","abstract_has_math":false,"creators":["Springer, Luca"],"institution":"Universität Potsdam","degree_name":null,"degree_level":"master","degree_discipline":null,"degree_department":null,"school":null,"contributors":["Tolovski, Ilin","Rabl, Tilmann","Polze, Andreas"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2024,"date_issued":"2024-11-25","date_published":"2024-11-25","updated_at":"2026-07-24T03:52:13Z","subjects":["carbon-aware","scheduling","slurm","temporal","spatial","shifting","kohlenstoffbewusst","Zeitplanerstellung","Verschiebung","räumlich","zeitlich"],"languages":[],"rights":["CC-BY - Namensnennung 4.0 International"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://publishup.uni-potsdam.de/frontdoor/index/index/docId/67320","outbound_label":"Repository record","outbound_source":"source_url"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Tolovski, Ilin","Rabl, Tilmann","Polze, Andreas"]},{"key":"dc:creator","label":"Author","values":["Springer, Luca"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:publisher","label":"Institution","values":["Universität Potsdam"]},{"key":"dc:type","label":"Dc Type","values":["masterThesis"]},{"key":"thesis:degree_level","label":"Degree Level","values":["master"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Universität Potsdam"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["carbon-aware","scheduling","slurm","temporal","spatial","shifting","kohlenstoffbewusst","Zeitplanerstellung","Verschiebung","räumlich","zeitlich"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["CC-BY - Namensnennung 4.0 International"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Despite significant advances in hardware efficiency, data centers continue increasing their electricity consumption due to rising application demand. Consequently, data center carbon emissions grow. In addition to energy-saving mechanisms, increasing progress in carbon-aware software is necessary to reduce carbon emissions and mitigate climate change. Enhancing workload managers used in data centers with carbon-aware mechanisms counteracts their emissions. State-of-the-art solutions do not cover carbon-aware scheduling for the widely used workload manager Slurm. This thesis develops and evaluates three carbon-aware scheduling algorithms for high-performance computing clusters using the Slurm scheduler. To this end, we elaborate on whether we can reduce operational emissions using spatiotemporal workload shifting within a Slurm cluster in Germany and how the effects differ across energy zones with different characteristics of their electricity grid. We develop and evaluate Squirrel, a carbon-aware batch job scheduler prototype that considers the grid carbon intensity and servers’ thermal design power values. We also propose a forecasting algorithm for grid carbon intensity that can be used when there is no access to forecasting data. We measure the energy profiles of different TPCx-AI jobs and use their hourly energy consumptions to simulate workload scheduling for three different cluster configurations. Our simulations compare Squirrel with carbon-agnostic first-in-first-out scheduling in five different energy zones. The comparison is based on the hourly computation of schedules using data throughout 2023. Squirrel achieves average emission savings between 0.3% and 32% in our simulations, demonstrating the practical implications of our research.","Trotz erheblicher Fortschritte bei der Hardware-Effizienz steigt der Stromverbrauch von Rechenzentren aufgrund der zunehmenden Anwendungsnachfrage weiter an. Folglich nehmen auch die Treibhausgasemissionen von Rechenzentren zu. Zusätzlich zu Energiesparmaßnahmen sind zunehmende Fortschritte bei der Entwicklung emissionsbewusster Software notwendig, um die Treibhausgasemissionen zu reduzieren und den Klimawandel abzuschwächen. Die Erweiterung der in Rechenzentren eingesetzten Workload-Manager um emissionsbewusste Mechanismen wirkt deren Emissionen entgegen. Der Stand der Technik deckt kein emissionsbewusstes Scheduling für den weit verbreiteten Workload-Manager Slurm ab. Diese Arbeit entwickelt und evaluiert drei Algorithmen für klimaschonendes Scheduling von High-Performance-Computing-Clustern, welche durch Slurm verwaltet werden. Zu diesem Zweck untersuchen wir, ob wir die betrieblichen Emissionen durch eine räumlich-zeitliche Verschiebung der Arbeitslast innerhalb eines Slurm-Clusters in Deutschland reduzieren können und wie sich die Auswirkungen in verschiedenen Energiezonen mit unterschiedlichen Eigenschaften des Stromnetzes unterscheiden. Wir entwickeln und evaluieren Squirrel, einen Prototyp eines emissionsbewussten Batch-Job-Schedulers, der die Emissionsintensität des Netzes und die thermische Verlustleistung der Server berücksichtigt. Wir schlagen auch einen Prognosealgorithmus für die Emissionsintensität des Netzes vor, der verwendet werden kann, wenn kein Zugang zu Prognosedaten besteht. Wir messen die Energieprofile verschiedener TPCx-AI-Jobs und verwenden ihren stündlichen Energieverbrauch, um die Arbeitslastplanung für drei verschiedene Clusterkonfigurationen zu simulieren. Unsere Simulationen vergleichen Squirrel mit Slurms First-in-First-out-Scheduling in fünf verschiedenen Energiezonen. Der Vergleich basiert auf der Berechnung von stündlichen Arbeitslastplanungen mit Daten aus dem Jahr 2023. Squirrel erzielt in unseren Simulationen durchschnittliche Emissionseinsparungen von 0,3% bis zu 32%, was die praktische Anwendbarkeit unserer Forschung zeigt."]},{"key":"dc:format.medium","label":"Dc Format Medium","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Squirrel"]}]}],"canonical_facts":{"dc:contributor":["Tolovski, Ilin","Rabl, Tilmann","Polze, Andreas"],"dc:creator":["Springer, Luca"],"dc:description.abstract":["Despite significant advances in hardware efficiency, data centers continue increasing their electricity consumption due to rising application demand. Consequently, data center carbon emissions grow. In addition to energy-saving mechanisms, increasing progress in carbon-aware software is necessary to reduce carbon emissions and mitigate climate change. Enhancing workload managers used in data centers with carbon-aware mechanisms counteracts their emissions. State-of-the-art solutions do not cover carbon-aware scheduling for the widely used workload manager Slurm. This thesis develops and evaluates three carbon-aware scheduling algorithms for high-performance computing clusters using the Slurm scheduler. To this end, we elaborate on whether we can reduce operational emissions using spatiotemporal workload shifting within a Slurm cluster in Germany and how the effects differ across energy zones with different characteristics of their electricity grid. We develop and evaluate Squirrel, a carbon-aware batch job scheduler prototype that considers the grid carbon intensity and servers’ thermal design power values. We also propose a forecasting algorithm for grid carbon intensity that can be used when there is no access to forecasting data. We measure the energy profiles of different TPCx-AI jobs and use their hourly energy consumptions to simulate workload scheduling for three different cluster configurations. Our simulations compare Squirrel with carbon-agnostic first-in-first-out scheduling in five different energy zones. The comparison is based on the hourly computation of schedules using data throughout 2023. Squirrel achieves average emission savings between 0.3% and 32% in our simulations, demonstrating the practical implications of our research.","Trotz erheblicher Fortschritte bei der Hardware-Effizienz steigt der Stromverbrauch von Rechenzentren aufgrund der zunehmenden Anwendungsnachfrage weiter an. Folglich nehmen auch die Treibhausgasemissionen von Rechenzentren zu. Zusätzlich zu Energiesparmaßnahmen sind zunehmende Fortschritte bei der Entwicklung emissionsbewusster Software notwendig, um die Treibhausgasemissionen zu reduzieren und den Klimawandel abzuschwächen. Die Erweiterung der in Rechenzentren eingesetzten Workload-Manager um emissionsbewusste Mechanismen wirkt deren Emissionen entgegen. Der Stand der Technik deckt kein emissionsbewusstes Scheduling für den weit verbreiteten Workload-Manager Slurm ab. Diese Arbeit entwickelt und evaluiert drei Algorithmen für klimaschonendes Scheduling von High-Performance-Computing-Clustern, welche durch Slurm verwaltet werden. Zu diesem Zweck untersuchen wir, ob wir die betrieblichen Emissionen durch eine räumlich-zeitliche Verschiebung der Arbeitslast innerhalb eines Slurm-Clusters in Deutschland reduzieren können und wie sich die Auswirkungen in verschiedenen Energiezonen mit unterschiedlichen Eigenschaften des Stromnetzes unterscheiden. Wir entwickeln und evaluieren Squirrel, einen Prototyp eines emissionsbewussten Batch-Job-Schedulers, der die Emissionsintensität des Netzes und die thermische Verlustleistung der Server berücksichtigt. Wir schlagen auch einen Prognosealgorithmus für die Emissionsintensität des Netzes vor, der verwendet werden kann, wenn kein Zugang zu Prognosedaten besteht. Wir messen die Energieprofile verschiedener TPCx-AI-Jobs und verwenden ihren stündlichen Energieverbrauch, um die Arbeitslastplanung für drei verschiedene Clusterkonfigurationen zu simulieren. Unsere Simulationen vergleichen Squirrel mit Slurms First-in-First-out-Scheduling in fünf verschiedenen Energiezonen. Der Vergleich basiert auf der Berechnung von stündlichen Arbeitslastplanungen mit Daten aus dem Jahr 2023. Squirrel erzielt in unseren Simulationen durchschnittliche Emissionseinsparungen von 0,3% bis zu 32%, was die praktische Anwendbarkeit unserer Forschung zeigt."],"dc:format.medium":["application/pdf"],"dc:publisher":["Universität Potsdam"],"dc:rights":["CC-BY - Namensnennung 4.0 International"],"dc:subject":["carbon-aware","scheduling","slurm","temporal","spatial","shifting","kohlenstoffbewusst","Zeitplanerstellung","Verschiebung","räumlich","zeitlich"],"dc:title":["Squirrel"],"dc:type":["masterThesis"],"thesis:degree_level":["master"],"thesis:institution_name":["Universität Potsdam"]},"updated_at":"2026-07-24T03:52:13Z"}