Carbon-Aware Scheduling on Multiple Edge Servers
複数エッジサーバーにおけるカーボンアウェアスケジューリング (AI 翻訳)
Joachim Cendrier, Rajini Wijayawardana, Anne Benoit, Yves Robert, Andrew A. Chien, Frédéric Vivien
🤖 gxceed AI 要約
日本語
本論文は、エッジサーバー群において炭素強度を考慮したジョブスケジューリングアルゴリズムを提案する。実際のデータと炭素トレースを用いたシミュレーションにより、標準手法と比較して平均42%の排出削減を達成した。
English
This paper presents carbon-aware scheduling algorithms for edge servers with varying carbon intensity. Using real-world job data and carbon traces, the proposed methods achieve an average 42% reduction in carbon emissions compared to standard approaches.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本ではデータセンターの省エネ・脱炭素が急務。エッジコンピューティングの普及に伴い、本手法は日本のクラウド事業者や通信キャリアのScope 2削減に貢献する可能性がある。
In the global GX context
As edge computing grows globally, carbon-aware scheduling becomes crucial for reducing operational emissions. This work provides algorithmic foundations applicable to any multi-server edge platform.
👥 読者別の含意
🔬研究者:This paper contributes scheduling algorithms with theoretical guarantees and practical gains for carbon-aware computing.
🏢実務担当者:Data center operators can apply these scheduling policies to reduce carbon footprint of edge servers.
📄 Abstract(原文)
<div> This work focuses on carbon-aware scheduling algorithms on edge platforms. We adopt a very general setting: (i) the edge servers have different carbon intensity levels, which are predicted more or less accurately through some horizon; (ii) jobs are submitted online with release times and deadlines; (iii) jobs can be preempted and migrated across edge servers at some nonzero cost. This work is the first attempt to introduce efficient algorithms to minimize total carbon emissions using such a complicated but realistic framework. We derive new complexity results and design several new algorithms that use sophisticated scheduling policies to efficiently decrease carbon cost. The more complicated algorithms re-evaluate all scheduling decisions to accommodate newly released jobs. We provide a comprehensive simulation campaign based on actual platform/job data and carbon traces and report an average gain of 42% over standard approaches. </div>
🔗 Provenance — このレコードを発見したソース
- openalex https://inria.hal.science/hal-05668204first seen 2026-06-29 05:35:35 · last seen 2026-06-29 05:35:42
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録(無料・3キーワードまで)。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。