publications | Abel Souza

2024

SoCC

TailClipper: Reducing Tail Response Time of Distributed Services Through System-Wide Scheduling

Nathan Ng, Abel Souza, Ahmed Ali-Eldin, David Irwin, Don Towsley, and Prashant Shenoy

In Proceedings of the 2024 ACM Symposium on Cloud Computing, 2024

Abs HTML PDF

Reducing tail latency has become a crucial issue for optimizing the performance of online cloud services and distributed applications. In distributed applications, there are many causes of high end-to-end tail latency, including operating system delays, request re-ordering due to fan-out/fanin, and network congestion. Although recent research has focused on reducing tail latency for individual application components, such as by replicating requests and scheduling, in this paper, we argue for a holistic approach for reducing the end-to-end tail latency across application components. We propose TailClipper, a distributed scheduler that tags each arriving request with an arrival timestamp, and propagates it across the microservices’ call chain. TailClipper then uses arrival timestamps to implement an oldest request first scheduler that combines global first-come first serve with a limited form of processor sharing to reduce end-to-end tail latency. In doing so, TailClipper can counter the performance degradation caused by request reordering in multi-tiered and microservices-based applications. We implement TailClipper as a userspace Linux scheduler and evaluate it using cloud workload traces and a real-world microservices application. Compared to state-of-the-art schedulers, our experiments reveal that TailClipper improves the 99th percentile response time by up to 81%, while also improving the mean response time and the system throughput by up to 54% and 29% respectively under high loads.
IGSC

PADS: Power Budgeting with Diagonal Scaling for Performance-Aware Cloud Workloads

Mehmet Savasci, Abel Souza, David Irwin, Ahmed Ali-Eldin, and Prashant Shenoy

In 2024 IEEE 15th International Green and Sustainable Computing Conference, 2024

Abs HTML PDF

Cloud platforms’ rapid growth raises significant concerns about their electricity consumption and resulting carbon emissions. Power capping is a known technique for limiting the power consumption of data centers where workloads are hosted. Today’s data center computer clusters co-locate latency-sensitive web and throughput-oriented batch workloads. When power capping is necessary, throttling only the batch tasks without restricting latency-sensitive web workloads is ideal because guaranteeing low response time for latency-sensitive workloads is a must due to Service-Level Objectives (SLOs) requirements. This paper proposes PADS, a hardware-agnostic workload-aware power capping system. Due to not relying on any hardware mechanism such as RAPL and DVFS, it can keep the power consumption of clusters equipped with heterogeneous architectures such as x86 and ARM below the enforced power limit while minimizing the impact on latency-sensitive tasks. It uses an application-performance model of both latency-sensitive and batch workloads to ensure power safety with controllable performance. Our power capping technique uses diagonal scaling and relies on using the control group feature of the Linux kernel. Our results indicate that PADS is highly effective in reducing power while respecting the tail latency requirement of the latency-sensitive workload. Furthermore, compared to state-of-the-art solutions, PADS demonstrates lower P95 latency, accompanied by a 90% higher effectiveness in respecting power limits.
BuildSys

Examining The Adoption of Electromobility Concepts Across Social Contexts For Energy Transition

Julia Köhlke, Adam Lechowicz, Oluwole Fabikun, Noman Bashir, Abel Souza, Prashant Shenoy, and Sebastian Lehnhoff

In Proceedings of the 11th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, 2024

HTML PDF
MILCOM

Enhancing Resilience in Distributed ML Inference Pipelines for Edge Computing

Li Wu, Walid A Hanafy, Abel Souza, Tarek Abdelzaher, Gunjan Verma, and Prashant Shenoy

In 2024-2024 IEEE Military Communications Conference, 2024

HTML PDF
MILCOM

Collaborative Inference in Resource-Constrained Edge Networks: Challenges and Opportunities

Nathan Ng, Abel Souza, Suhas Diggavi, Niranjan Suri, Tarek Abdelzaher, Don Towsley, and Prashant Shenoy

In MILCOM 2024-2024 IEEE Military Communications Conference, 2024

HTML PDF
ICCCN

Acies-OS: A Content-Centric Platform for Edge AI Twinning and Orchestration

Jinyang Li, Yizhuo Chen, Tomoyoshi Kimura, Tianshi Wang, Ruijie Wang, Denizhan Kara, Yigong Hu, Li Wu, Walid A Hanafy, Abel Souza, and others

In 2024 33rd International Conference on Computer Communications and Networks, 2024

HTML
ASPLOS

Going Green for Less Green: Optimizing the Cost of Reducing Cloud Carbon Emissions

Walid A. Hanafy, Qianlin Liang, Noman Bashir, Abel Souza, David Irwin, and Prashant Shenoy

In Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3, 2024

Abs HTML

The continued exponential growth of cloud datacenter capacity has increased awareness of the carbon emissions when executing large compute-intensive workloads. To reduce carbon emissions, cloud users often temporally shift their batch workloads to periods with low carbon intensity. While such time shifting can increase job completion times due to their delayed execution, the cost savings from cloud purchase options, such as reserved instances, also decrease when users operate in a carbon-aware manner. This happens because carbon-aware adjustments change the demand pattern by periodically leaving resources idle, which creates a trade-off between carbon emissions and cost. In this paper, we present GAIA, a carbon-aware scheduler that enables users to address the three-way trade-off between carbon, performance, and cost in cloud-based batch schedulers. Our results quantify the carbon-performance-cost trade-off in cloud platforms and show that compared to existing carbon-aware scheduling policies, our proposed policies can double the amount of carbon savings per percentage increase in cost, while decreasing the performance overhead by 26%.

e-Energy

On the Implications of Choosing Average versus Marginal Carbon Intensity Signals on Carbon-aware Optimizations

Thanathorn Sukprasert, Noman Bashir, Abel Souza, David Irwin, and Prashant Shenoy

In Proceedings of the 15th ACM International Conference on Future and Sustainable Energy Systems (e-Energy), 2024

Bib

@inproceedings{sukprasert2024signals,
  title = {On the Implications of Choosing Average versus Marginal Carbon Intensity Signals on Carbon-aware Optimizations},
  author = {Sukprasert, Thanathorn and Bashir, Noman and Souza, Abel and Irwin, David and Shenoy, Prashant},
  year = {2024},
  booktitle = {Proceedings of the 15th ACM International Conference on Future and Sustainable Energy Systems (e-Energy)}
}

CCGrid

SLO-Power: SLO and Power-aware Elastic Scaling for Web Sevices

Mehmet Savasci, Abel Souza, Li Wu, David Irwin, Ahmed Ali-Eldin, and Prashant Shenoy

In Proceedings of the 24th IEEE/ACM International Symposium on Cluster, Cloud, and Internet Computing (CCGrid), 2024

Bib PDF

@inproceedings{savasci2024slopower,
  title = {SLO-Power: SLO and Power-aware Elastic Scaling for Web Sevices},
  author = {Savasci, Mehmet and Souza, Abel and Wu, Li and Irwin, David and Ali-Eldin, Ahmed and Shenoy, Prashant},
  year = {2024},
  booktitle = {Proceedings of the 24th IEEE/ACM International Symposium on Cluster, Cloud, and Internet Computing (CCGrid)}
}

EuroSys
On the Limitations of Carbon-Aware Temporal and Spatial Workload Shifting in the Cloud

Thanathorn Sukprasert, Abel Souza, Noman Bashir, David Irwin, and Prashant Shenoy

In Proceedings of the 19th European Conference on Computer Systems (EuroSys), 2024

Abs arXiv Bib

Cloud platforms have been focusing on reducing their carbon emissions by shifting workloads across time and locations to when and where low-carbon energy is available. Despite the prominence of this idea, prior work has only quantified the potential of spatiotemporal workload shifting in narrow settings, i.e., for specific workloads in select regions. In particular, there has been limited work on quantifying an upper bound on the ideal and practical benefits of carbon-aware spatiotemporal workload shifting for a wide range of cloud workloads. To address the problem, we conduct a detailed data-driven analysis to understand the benefits and limitations of carbon-aware spatiotemporal scheduling for cloud workloads. We utilize carbon intensity data from 123 regions, encompassing most major cloud sites, to analyze two broad classes of workloads – batch and interactive – and their various characteristics, e.g., job duration, deadlines, and SLOs. Our findings show that while spatiotemporal workload shifting can reduce workloads’ carbon emissions, the practical upper bounds of these carbon reductions are currently limited and far from ideal. We also show that simple scheduling policies often yield most of these reductions, with more sophisticated techniques yielding little additional benefit. Notably, we also find that the benefit of carbon-aware workload scheduling relative to carbon-agnostic scheduling will decrease as the energy supply becomes “greener”.
@inproceedings{sukprasert2024limitations, title = {On the Limitations of Carbon-Aware Temporal and Spatial Workload Shifting in the Cloud}, author = {Sukprasert, Thanathorn and Souza, Abel and Bashir, Noman and Irwin, David and Shenoy, Prashant}, year = {2024}, booktitle = {Proceedings of the 19th European Conference on Computer Systems (EuroSys)}, }

2023

ASPLOS

Ecovisor: A virtual energy system for carbon-efficient applications

Abel Souza, Noman Bashir, Jorge Murillo, Walid Hanafy, Qianlin Liang, David Irwin, and Prashant Shenoy

In Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2, 2023

arXiv Bib HTML PDF Code Slides

@inproceedings{ecovisor,
  title = {Ecovisor: A virtual energy system for carbon-efficient applications},
  author = {Souza, Abel and Bashir, Noman and Murillo, Jorge and Hanafy, Walid and Liang, Qianlin and Irwin, David and Shenoy, Prashant},
  booktitle = {Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2},
  pages = {252--265},
  year = {2023},
  doi = {10.1145/3575693.3575709},
}

IGSC

CASPER: Carbon-Aware Scheduling and Provisioning for Distributed Web Services

Abel Souza, Jasoria Shruti, Basundhara Chakrabarty, Alexander Bridgwater, Axel Lundberg, Filip Skogh, Ahmed Ali-Eldin, David Irwin, and Prashant Shenoy

In Proceedings of the 14th International Green and Sustainable Computing Conference (IGSC), Toronto, ON, Canada, 2023

Bib PDF

@inproceedings{casper,
  title = {CASPER: Carbon-Aware Scheduling and Provisioning for Distributed Web Services},
  author = {Souza, Abel and Shruti, Jasoria and Chakrabarty, Basundhara and Bridgwater, Alexander and Lundberg, Axel and Skogh, Filip and Ali-Eldin, Ahmed and Irwin, David and Shenoy, Prashant},
  booktitle = {Proceedings of the 14th International Green and Sustainable Computing Conference (IGSC), Toronto, ON, Canada},
  year = {2023},
}

BuildSys

Empowering User-Centered Carbon Management: Bridging Individual Preferences and Sociotechnical Advancements

Abel Souza, Mihir Shenoy, and Camellia Zakaria

In Proceedings of the 10th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, 2023

Bib HTML PDF Slides

@inproceedings{BuildSys.souza2023empowering,
  title = {Empowering User-Centered Carbon Management: Bridging Individual Preferences and Sociotechnical Advancements},
  author = {Souza, Abel and Shenoy, Mihir and Zakaria, Camellia},
  booktitle = {Proceedings of the 10th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation},
  pages = {406--412},
  year = {2023},
}

e-Energy

Spatiotemporal Carbon-aware Scheduling in the Cloud: Limits and Benefits

Thanathorn Sukprasert, Abel Souza, Noman Bashir, David Irwin, and Prashant Shenoy

2023

Bib HTML

@article{sukprasert2023quantifying,
  title = {Spatiotemporal Carbon-aware Scheduling in the Cloud: Limits and Benefits},
  author = {Sukprasert, Thanathorn and Souza, Abel and Bashir, Noman and Irwin, David and Shenoy, Prashant},
  year = {2023}
}

MILCOM

Unlocking Efficiency: Understanding End-to-End Performance in Distributed Analytics Pipelines

Abel Souza, Nathan Ng, Tarek Abdelzaher, Don Towsley, and Prashant Shenoy

In Proceedings of the 41st IEEE Military Communications Conference, Workshop on Internet of Things for Adversarial Environments, 2023

Bib PDF

@inproceedings{souzaunlocking,
  title = {Unlocking Efficiency: Understanding End-to-End Performance in Distributed Analytics Pipelines},
  author = {Souza, Abel and Ng, Nathan and Abdelzaher, Tarek and Towsley, Don and Shenoy, Prashant},
  booktitle = {Proceedings of the 41st IEEE Military Communications Conference, Workshop on Internet of Things for Adversarial Environments},
  year = {2023}
}

2022

HotCarbon

Sustainable computing-without the hot air

Noman Bashir, David Irwin, Prashant Shenoy, and Abel Souza

ACM SIGENERGY Energy Informatics Review, 2022

arXiv Bib HTML

@article{bashir2023sustainable,
  title = {Sustainable computing-without the hot air},
  author = {Bashir, Noman and Irwin, David and Shenoy, Prashant and Souza, Abel},
  journal = {ACM SIGENERGY Energy Informatics Review},
  volume = {3},
  number = {3},
  pages = {47--52},
  year = {2022},
  publisher = {ACM New York, NY, USA}
}

Universe

Launching the VASCO citizen science project

Beatriz Villarroel, Kristiaan Pelckmans, Enrique Solano, Mikael Laaksoharju, Abel Souza, Onyeuwaoma Nnaemeka Dom, Khaoula Laggoune, Jamal Mimouni, Hichem Guergouri, Lars Mattsson, and others

Universe, 2022

Bib HTML PDF

@article{villarroel2022launching,
  title = {Launching the VASCO citizen science project},
  author = {Villarroel, Beatriz and Pelckmans, Kristiaan and Solano, Enrique and Laaksoharju, Mikael and Souza, Abel and Dom, Onyeuwaoma Nnaemeka and Laggoune, Khaoula and Mimouni, Jamal and Guergouri, Hichem and Mattsson, Lars and others},
  journal = {Universe},
  volume = {8},
  number = {11},
  pages = {561},
  year = {2022},
  publisher = {MDPI},
  doi = {10.3390/universe8110561},
  url = {https://doi.org/10.3390/universe8110561},
  dimensions = {true}
}

APpLIED

DARTS: Distributed IoT Architecture for Real-Time, Resilient and AI-Compressed Workflows

Ragini Gupta, Bo Chen, Shengzhong Liu, Tianshi Wang, Sandeep Singh Sandha, Abel Souza, Klara Nahrstedt, Tarek Abdelzaher, Mani Srivastava, Prashant Shenoy, and others

In Proceedings of the 2022 Workshop on Advanced tools, programming languages, and PLatforms for Implementing and Evaluating algorithms for Distributed systems, 2022

Bib HTML PDF

@inproceedings{gupta2022darts,
  title = {DARTS: Distributed IoT Architecture for Real-Time, Resilient and AI-Compressed Workflows},
  author = {Gupta, Ragini and Chen, Bo and Liu, Shengzhong and Wang, Tianshi and Sandha, Sandeep Singh and Souza, Abel and Nahrstedt, Klara and Abdelzaher, Tarek and Srivastava, Mani and Shenoy, Prashant and others},
  booktitle = {Proceedings of the 2022 Workshop on Advanced tools, programming languages, and PLatforms for Implementing and Evaluating algorithms for Distributed systems},
  pages = {15--23},
  year = {2022}
}

2021

JSSPP

A HPC Co-Scheduler with Reinforcement Learning

Abel Souza, Kristiaan Pelckmans, and Johan Tordsson

In Job Scheduling Strategies for Parallel Processing: 24th International Workshop, JSSPP 2021, Virtual Event, May 21, 2021, Revised Selected Papers 24, 2021

arXiv Bib HTML PDF Video Slides

@inproceedings{souza2021hpc,
  presentation = {https://youtu.be/69syWGxH_mk?si=HOLqz26VkGZsEGyC&t=1743},
  title = {A HPC Co-Scheduler with Reinforcement Learning},
  author = {Souza, Abel and Pelckmans, Kristiaan and Tordsson, Johan},
  booktitle = {Job Scheduling Strategies for Parallel Processing: 24th International Workshop, JSSPP 2021, Virtual Event, May 21, 2021, Revised Selected Papers 24},
  pages = {126--148},
  year = {2021},
  organization = {Springer}
}

SoCC

Enabling sustainable clouds: The case for virtualizing the energy system

Noman Bashir, Tian Guo, Mohammad Hajiesmaili, David Irwin, Prashant Shenoy, Ramesh Sitaraman, Abel Souza, and Adam Wierman

In Proceedings of the ACM Symposium on Cloud Computing, 2021

Bib HTML PDF Video

@inproceedings{bashir2021enabling,
  presentation = {https://youtu.be/m0RLpmOCpv4?si=en2plapYoPNcP_-5},
  title = {Enabling sustainable clouds: The case for virtualizing the energy system},
  author = {Bashir, Noman and Guo, Tian and Hajiesmaili, Mohammad and Irwin, David and Shenoy, Prashant and Sitaraman, Ramesh and Souza, Abel and Wierman, Adam},
  booktitle = {Proceedings of the ACM Symposium on Cloud Computing},
  pages = {350--358},
  year = {2021},
}

EDBT

AdCom: Adaptive Combiner for Streaming Aggregations

Felipe Gutierrez, Kaustubh Beedkar, Abel Souza, and Volker Markl

In EDBT 2021-24th International Conference on Extending Database Technology, 2021

Bib PDF Code Video

@inproceedings{gutierrez2021adcom,
  title = {AdCom: Adaptive Combiner for Streaming Aggregations},
  author = {Gutierrez, Felipe and Beedkar, Kaustubh and Souza, Abel and Markl, Volker},
  booktitle = {EDBT 2021-24th International Conference on Extending Database Technology},
  year = {2021},
  presentation = {https://youtu.be/fWF0b51vA9Y?si=jUuPus4BsHCpXytO},
}

2020

HPDC

ASA – The Adaptive Scheduling Architecture

Abel Souza, Kristiaan Pelckmans, Devarshi Ghoshal, Lavanya Ramakrishnan, and Johan Tordsson

In Proceedings of the 29th International Symposium on High-Performance Parallel and Distributed Computing, 2020

arXiv Bib HTML PDF Video

@inproceedings{souza2020asa,
  title = {ASA -- The Adaptive Scheduling Architecture},
  author = {Souza, Abel and Pelckmans, Kristiaan and Ghoshal, Devarshi and Ramakrishnan, Lavanya and Tordsson, Johan},
  booktitle = {Proceedings of the 29th International Symposium on High-Performance Parallel and Distributed Computing},
  pages = {161--165},
  year = {2020},
  presentation = {https://youtu.be/r9r1SwJH-wM?si=qMd6OOYYx05zpqMr}
}

PhD Thesis

Autonomous resource management for high performance datacenters

Abel Souza

Department of computing science, Umeå University, 2020

Bib HTML Video

@phdthesis{souza2020autonomous,
  publisher = {PhD},
  title = {Autonomous resource management for high performance datacenters},
  author = {Souza, Abel},
  year = {2020},
  school = {Department of computing science, Ume{\aa} University},
  presentation = {https://archive.cs.umu.se/phd-defence/AbelSouza-2020-05-08/},
}

2019

CCGrid

Hybrid Resource Management for HPC and Data Intensive Workloads

Abel Souza, Mohamad Rezaei, Erwin Laure, and Johan Tordsson

In 2019 19th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID), 2019

Bib HTML

@inproceedings{souza2019hybrid,
  title = {Hybrid Resource Management for HPC and Data Intensive Workloads},
  author = {Souza, Abel and Rezaei, Mohamad and Laure, Erwin and Tordsson, Johan},
  booktitle = {2019 19th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)},
  pages = {399--409},
  year = {2019},
  organization = {IEEE}
}

2018

IC2E

Hybrid adaptive checkpointing for virtual machine fault tolerance

Abel Souza, Alessandro Vittorio Papadopoulos, Luis Tomas, David Gilbert, and Johan Tordsson

In 2018 IEEE International Conference on Cloud Engineering (IC2E), 2018

Bib HTML Slides

@inproceedings{souza2018hybrid,
  title = {Hybrid adaptive checkpointing for virtual machine fault tolerance},
  author = {Souza, Abel and Papadopoulos, Alessandro Vittorio and Tomas, Luis and Gilbert, David and Tordsson, Johan},
  booktitle = {2018 IEEE International Conference on Cloud Engineering (IC2E)},
  pages = {12--22},
  year = {2018},
  organization = {IEEE}
}

Licentiate Thesis

Application-aware resource management for datacenters

Abel Souza

Department of computing science, Umeå university, 2018

Bib HTML

@phdthesis{souza2018application,
  publisher = {PhLic},
  title = {Application-aware resource management for datacenters},
  author = {Souza, Abel},
  year = {2018},
  school = {Department of computing science, Ume{\aa} university},
}

2017

WORKS

E-HPC: A Library for Elastic Resource Management in HPC Environments

William Fox, Devarshi Ghoshal, Abel Souza, Gonzalo P Rodrigo, and Lavanya Ramakrishnan

In Proceedings of the 12th Workshop on Workflows in Support of Large-Scale Science, 2017

Bib HTML PDF

@inproceedings{fox2017hpc,
  title = {E-HPC: A Library for Elastic Resource Management in HPC Environments},
  author = {Fox, William and Ghoshal, Devarshi and Souza, Abel and Rodrigo, Gonzalo P and Ramakrishnan, Lavanya},
  booktitle = {Proceedings of the 12th Workshop on Workflows in Support of Large-Scale Science},
  pages = {1--11},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  year = {2017}
}