The AI datacenters are currently being deployed on a large scale to support the training and deployment of power-intensive large-language models (LLMs). Extensive amount of computation and cooling required in datacenters increase concerns about the energy use and carbon emissions of AI datacenters. Although current state-of-the-art has examined the energy efficiency of LLM inference, most prior research focused on optimizing compute-side scheduling without considering thermal objectives or constraints. Since GPU-intensive inference generates substantial heat that can degrade datacenter performance, ignoring thermal effects can increase total energy consumption and reduce the efficiency of LLM serving. To fill this gap, we profile the characteristics of GPU servers under varying cooling and AI jobs, and develop a joint cooling and computing modeling approach for AI datacenters. Built upon such workload and thermal dynamics models, a novel hierarchical control framework is proposed to co-optimize computing and thermal management by identifying the optimal GPU parallelism, frequency (DVFS), and cooling control knobs. Using real Azure inference traces and detailed GPU profiling, our mo
Recent proposals for datacenters in sun-synchronous Low Earth Orbit rely on a large number of compute satellites formation-flying in dense clusters. Designing such satellite clusters requires optimizing the satellites' orbital geometry under several safety and operational constraints applied throughout the cluster's entire orbit. These constraints include guaranteeing a minimum inter-satellite spacing, obstruction-less solar power for every satellite, and that each satellite have a stable set of nearest neighbors with which it can maintain inter-satellite links (ISLs). In this work, we propose two main cluster orbital designs, parametrized by the minimum inter-satellite spacing $R_{min}$ and the cluster radius $R_{max}$: a planar cluster, and a 3D cluster. We show by construction and numerical analysis that both cluster orbital designs are consistent with the inter-satellite spacing, unobstructed sun-vector, and inter-satellite line of sight constraints. The proposed planar architecture is the most efficient packing of satellites in a plane for given $R_{min}$ and $R_{max}$ values, and our 3D architecture allows for the number of datacenter satellites to scale proportional to $(R_{
Organizations are increasingly moving towards the cloud computing paradigm, in which an on-demand access to a pool of shared configurable resources is provided. However, security challenges, which are particularly exacerbated by the multitenancy and virtualization features of cloud computing, present a major obstacle. In particular, sharing of resources among potentially untrusted tenants in access controlled cloud datacenters can result in increased risk of data leakage. To address such risk, we propose an efficient risk-aware sensitive property-driven virtual resource assignment mechanism for cloud datacenters. We have used two information-theoretic measures, i.e., KL-divergence and mutual information, to represent sensitive properties in the dataset. Based on the vulnerabilities of cloud architecture and the sensitive property profile, we have formulated the problem as a cost-drive optimization problem. The problem is shown to be NP-complete. Accordingly, we have proposed two heuristics and presented simulation based performance results for cloud datacenters with multiple sensitivity.
Datacenters have become the backbone of modern digital infrastructure, powering the rapid rise of artificial intelligence and promising economic growth and technological progress. However, this expansion has brought growing tensions in the local communities where datacenters are already situated or being proposed. While the mainstream discourse often focuses on energy usage and carbon footprint of the computing sector at a global scale, the local socio-environmental consequences -- such as health impacts, water usage, noise pollution, infrastructural strain, and economic burden -- remain largely underexplored and poorly addressed. In this work, we surface these community-level consequences through a mixed-methods study that combines quantitative data with qualitative insights. Focusing on Northern Virginia's ``Data Center Valley,'' we highlight how datacenter growth reshapes local environments and everyday life, and examine the power dynamics that determine who benefits and who bears the costs. Our goal is to bring visibility to these impacts and prompt more equitable and informed decisions about the future of digital infrastructure.
As the Middle East emerges as a strategic hub for artificial intelligence (AI) infrastructure, the feasibility of deploying sustainable datacenters in desert environments has become a topic of growing relevance. This paper presents an empirical study analyzing the energy consumption and carbon footprint of large language model (LLM) inference across four countries: the United Arab Emirates, Iceland, Germany, and the United States of America using DeepSeek Coder 1.3B and the HumanEval dataset on the task of code generation. We use the CodeCarbon library to track energy and carbon emissions andcompare geographical trade-offs for climate-aware AI deployment. Our findings highlight both the challenges and potential of datacenters in desert regions and provide a balanced outlook on their role in global AI expansion.
Recent trends see a move away from a fixed-resource server-centric datacenter model to a more adaptable "disaggregated" datacenter model. These disaggregated datacenters can then dynamically group resources to the specific requirements of an incoming workload, thereby improving efficiency. To properly utilize these disaggregated datacenters, workload allocation techniques must examine the current state of the datacenter and choose resources that not only optimize the current workload request, but future ones. Since disaggregated datacenters are severely bottlenecked by the available network resources, our work proposes a heuristic-based approach called RISA, which significantly reduces the network usage of workload allocations in disaggregated datacenters. Compared to the state-of-the-art, RISA reduces the power consumption for optical components by 33% and reduces the average CPU-RAM round-trip latency by 50%. Additionally, RISA significantly outperforms the state-of-the-art in terms of execution time.
Cloud providers are adapting datacenter (DC) capacity to reduce carbon emissions. With hyperscale datacenters exceeding 100 MW individually, and in some grids exceeding 15% of power load, DC adaptation is large enough to harm power grid dynamics, increasing carbon emissions, power prices, or reduce grid reliability. To avoid harm, we explore coordination of DC capacity change varying scope in space and time. In space, coordination scope spans a single datacenter, a group of datacenters, and datacenters with the grid. In time, scope ranges from online to day-ahead. We also consider what DC and grid information is used (e.g. real-time and day-ahead average carbon, power price, and compute backlog). For example, in our proposed PlanShare scheme, each datacenter uses day-ahead information to create a capacity plan and shares it, allowing global grid optimization (over all loads, over entire day). We evaluate DC carbon emissions reduction. Results show that local coordination scope fails to reduce carbon emissions significantly (3.2%--5.4% reduction). Expanding coordination scope to a set of datacenters improves slightly (4.9%--7.3%). PlanShare, with grid-wide coordination and full-day
Rail-optimized network fabrics have become the de facto datacenter scale-out fabric for large-scale ML training. However, the use of high-radix electrical switches to provide all-to-all connectivity in rails imposes massive power, cost, and complexity overheads. We propose a rethinking of the rail abstraction by retaining its communication semantics, but realizing it using optical circuit switches. The key challenge is that optical switches support only one-to-one connectivity at a time, limiting the fan-out of traffic in ML workloads using hybrid parallelisms. We introduce parallelism-driven rail reconfiguration as a solution that leverages the sequential ordering between traffic from different parallelisms. We design a control plane, Opus, to enable time-multiplexed emulation of electrical rail switches using optical switches. More broadly, our work discusses a new research agenda: datacenter fabrics that co-evolve with the model parallelism dimensions within each job, as opposed to the prevailing mindset of reconfiguring networks before a job begins.
We introduce Capsule, a mechanism for seamlessly sharing datacenter resources across multiple players. It decouples player-local and global states to achieve isolation and to maximize cross-player sharing. Our evaluations show that Capsule increases datacenter resource utilization by accommodating up to 2.25x more players without degrading the user experience. This improvement stems from Capsule consuming up to 1.43x less GPU, 3.11x less VRAM, 3.7x less CPU, and 3.87x less RAM compared to the baseline. We evaluated Capsule across four applications and various hardware configurations, including three distinct servers and a multi-server cluster. These results demonstrate that the Capsule design is portable to other game engines.
Cloud providers are highly incentivized to reduce latency. One way they do this is by locating datacenters as close to users as possible. These "cloud edge" datacenters are placed in metropolitan areas and enable edge computing for residents of these cities. Therefore, which cities are selected to host edge datacenters determines who has the fastest access to applications requiring edge compute - creating a digital divide between those closest and furthest from a datacenter. In this study we measure latency to the current and predicted cloud edge datacenters of three major cloud providers around the world. Our measurements use the RIPE Atlas platform targeting cloud regions, AWS Local Zones, and network optimization services that minimize the path to the cloud edge. An analysis of the digital divide shows rising inequality as the relative difference between users closest and farthest from cloud compute increases. We also find this inequality unfairly affects lower income census tracts in the US. This result is extended globally using remotely sensed night time lights as a proxy for wealth. Finally, we demonstrate that low earth orbit satellite internet can help to close this digita
The exponential growth in demand for digital services drives massive datacenter energy consumption and negative environmental impacts. Promoting sustainable solutions to pressing energy and digital infrastructure challenges is crucial. Several hyperscale cloud providers have announced plans to power their datacenters using renewable energy. However, integrating renewables to power the datacenters is challenging because the power generation is intermittent, necessitating approaches to tackle power supply variability. Hand engineering domain-specific heuristics-based schedulers to meet specific objective functions in such complex dynamic green datacenter environments is time-consuming, expensive, and requires extensive tuning by domain experts. The green datacenters need smart systems and system software to employ multiple renewable energy sources (wind and solar) by intelligently adapting computing to renewable energy generation. We present RARE (Renewable energy Aware REsource management), a Deep Reinforcement Learning (DRL) job scheduler that automatically learns effective job scheduling policies while continually adapting to datacenters' complex dynamic environment. The resulting
Amidst the climate crisis, the massive introduction of renewable energy sources has brought tremendous challenges to both the power grid and its surrounding markets. As datacenters have become ever-larger and more powerful, they play an increasingly significant role in the energy arena. With their unique characteristics, datacenters have been proved to be well-suited for regulating the power grid yet currently provide little, if any, such active response. This problem is due to issues such as unsuitability of the market design, high complexity of the currently proposed solutions, as well as the potential risks thereof. This work aims to provide individual datacenters with insights on the feasibility and profitability of directly participating in the energy market. By modelling the power system of datacenters, and by conducting simulations on real-world datacenter traces, we demonstrate the substantial financial incentive for individual datacenters to directly participate in both the day-ahead and the balancing markets. In turn, we suggest a new short-term, direct scheme of market participation for individual datacenters in place of the current long-term, inactive participation. Fur
Emerging reconfigurable datacenters allow to dynamically adjust the network topology in a demand-aware manner. These datacenters rely on optical switches which can be reconfigured to provide direct connectivity between racks, in the form of edge-disjoint matchings. While state-of-the-art optical switches in principle support microsecond reconfigurations, the demand-aware topology optimization constitutes a bottleneck. This paper proposes a dynamic algorithms approach to improve the performance of reconfigurable datacenter networks, by supporting faster reactions to changes in the traffic demand. This approach leverages the temporal locality of traffic patterns in order to update the interconnecting matchings incrementally, rather than recomputing them from scratch. In particular, we present six (batch-)dynamic algorithms and compare them to static ones. We conduct an extensive empirical evaluation on 176 synthetic and 39 real-world traces, and find that dynamic algorithms can both significantly improve the running time and reduce the number of changes to the configuration, especially in networks with high temporal locality, while retaining matching weight.
This paper represents the first effort to quantify uncertainty in carbon intensity forecasting for datacenter decarbonization. We identify and analyze two types of uncertainty -- temporal and spatial -- and discuss their system implications. To address the temporal dynamics in quantifying uncertainty for carbon intensity forecasting, we introduce a conformal prediction-based framework. Evaluation results show that our technique robustly achieves target coverages in uncertainty quantification across various significance levels. We conduct two case studies using production power traces, focusing on temporal and spatial load shifting respectively. The results show that incorporating uncertainty into scheduling decisions can prevent a 5% and 14% increase in carbon emissions, respectively. These percentages translate to an absolute reduction of 2.1 and 10.4 tons of carbon emissions in a 20 MW datacenter cluster.
Datacenters are the backbone of our digital society, but raise numerous operational challenges. We envision digital twins becoming primary instruments in datacenter operations, continuously and autonomously helping with major operational decisions and with adapting ICT infrastructure, live, with a human-in-the-loop. Although fields such as aviation and autonomous driving successfully employ digital twins, an open-source digital twin for datacenters has not been demonstrated to the community. Addressing this challenge, we design, implement, and experiment using OpenDT, an Open-source, Digital Twin for monitoring and operating datacenters through a continuous integration cycle that includes: (1) live and continuous telemetry data; (2) discrete-event simulation using live telemetry from the physical ICT, with self-calibration; and (3) SLO-aware and human-approved feedback to physical ICT. Through trace-driven experiments with a prototype mainly covering stages 1 and 2 of the cycle, we show that (i) OpenDT can be used to reproduce peer-reviewed experiments and extend the analysis with performance and energy-efficiency results; (ii) OpenDT's online re-calibration can increase digital-tw
Diversification of digital applications and workloads has driven the development of diverse datacenter architectures on ever-larger scales. These datacenters consist of complex IT, power, and cooling systems with interdependencies that influence configuration and performance. As datacenters scale and power density increase, designing realistic models becomes more difficult, particularly for research, because it requires understanding all layers of the datacenter and how they interact. Consequently, many studies rely on outdated or unrealistic designs. To support research in datacenter hardware design principles, operational dynamics, cooling mechanisms, and interactions of these facilities with the electrical grid, we have designed DCGen, a tool which can generate a variety of datacenter configurations (including IT hardware, cooling and power distribution infrastructures) at various electrical power, compute capability, and area targets.The tool captures power and space characteristics of IT, cooling, and power infrastructures at both the rack and datacenter levels, enabling modeling of power, energy, and space. DCGen leverages specific use cases such as AI training, AI inference,
Datacenters are vital to our digital society, but consume a considerable fraction of global electricity and demand is projected to increase. To improve their sustainability and performance, we envision that simulators will become primary decision-making tools. However, and unlike other fields focusing on key societal infrastructure such as waterworks and mass transit, datacenter simulators do not yet combine multiple independent models into their operation and thus suffer from issues associated with singular models, such as specialization, and lack of adaptability to operational phenomena. To address this challenge, we propose M3SA, a datacenter simulation and analysis framework that uses discrete-event simulation to predict, for each model, the impact on climate and performance under various realistic datacenter conditions, and then combines these predictions. We design an architecture for simulating multiple concurrent models (Multi-Model), a technique for integrating the results of multiple models into a Meta-Model, and a procedure for quantifying Meta-Model accuracy. Through experiments with an M3SA prototype, we show that (i) M3SA can reproduce and enhance peer-reviewed experi
Today, the rapid growth of applications reliant on datacenters calls for new advancements to meet the increasing traffic and computational demands. Traffic traces from datacenters are essential for further development and optimization of future datacenters. However, traces are rarely released to the public. Researchers often use simplified mathematical models that lack the depth needed to recreate intricate traffic patterns and, thus, miss optimization opportunities found in realistic traffic. In this preliminary work, we introduce DTG-GPT, a packet-level Datacenter Traffic Generator (DTG), based on the generative pre-trained transformer (GPT) architecture used by many state-of-the-art large language models. We train our model on a small set of available traffic traces from different domains and offer a simple methodology to evaluate the fidelity of the generated traces to their original counterparts. We show that DTG-GPT can synthesize novel traces that mimic the spatiotemporal patterns found in real traffic traces. We further demonstrate that DTG-GPT can generate traces for networks of different scales while maintaining fidelity. Our findings indicate the potential that, in the f
AI's growing compute demand and new datacenter buildouts present major capacity and reliability challenges for the electricity grid, leading to multi-year interconnection delays for new datacenters and bottlenecking AI growth. To ease this strain, datacenters increasingly offer rapid power flexibility in response to grid signals, where the datacenter can increase or decrease its power consumption by adapting its workload in real time. In order to understand the impact of large datacenters on the grid and to facilitate the design of effective coordination strategies, we build OpenG2G, a simulation platform for AI datacenter-grid runtime coordination. We show that OpenG2G is capable of answering a wide range of coordination questions by allowing users to implement and compare various control paradigms (including classic, optimization, and learning-based controllers), and quantify how AI model and deployment choices affect datacenter flexibility and coordination outcomes. This versatility is enabled by OpenG2G's modular and extensible architecture: a datacenter backend driven by real measurements of production-grade AI services, a grid backend built on high-fidelity grid simulators, a
This paper proposes a hybrid energy storage system (HESS)-based control framework that enables comprehensive power smoothing for hyperscale AI datacenters with large load variations. Datacenters impose severe ramping and fluctuation-induced stresses on the grid frequency and voltage stability. To mitigate such disturbances, the proposed HESS integrates a battery energy storage system (BESS) and a supercapacitor (SC) through coordinated multi-timescale control. A high-pass filter (HPF) separates the datacenter demand into slow and fast components, allocating them respectively to the ESS via a leaky-integral controller and to the SC via a phase-lead proportional-derivative controller enhanced with feedforward and ramp-tracking compensation. Adaptive weighting and repetitive control mechanisms further improve transient and periodic responses. Case studies verify that the proposed method effectively suppresses both ramping and fluctuations, stabilizes the system frequency, and maintains sustainable state-of-charge (SoC) trajectories for both ESS and SC under prolonged, stochastic training cycles.