SHF: Medium: WHOLEPRO: An Online, Holistic Job Scheduling and Resource Provisioning Framework for Datacenter Architectures and Applications

SHF：中：WHOLEPRO：用于数据中心架构和应用程序的在线整体作业调度和资源配置框架

• 批准号: 1704504
• 负责人: Hao Che
• 金额: $ 104.59万
• 依托单位: University of Texas at Arlington
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-06-01 至 2021-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1704504&HistoricalAwards=false
• 关键词: SHF; Medium; WHOLEPRO; Online; Holistic

The ultimate design goal of datacenter services is to enable user-centric services that meet diverse Quality-of-Experience (QoE) requirements for individual users, while allowing for optimized and fair computing and networking resource sharing among users. However, achieving this goal is by no means easy. Existing datacenter job scheduling and resource provisioning mechanisms are bottom-up, point solutions, addressing partial aspects of the related design challenges only, and hence, fail to achieve the above goal.       The proposed research will develop a top-down, holistic approach concerning the entire datacenter systems stack, aiming at achieving the dual goal. It starts from the very top, i.e., the QoE or user utility requirements at the user application level. It then works its way downward, by mapping user-level requirements into datacenter system-level resource demands, which, together with the datacenter computing and networking resource constraints, set the boundary conditions for a proposed user-utility-aware optimization framework. Finally, an example software-based design and implementation of the proposed solution will be developed for a state-of-the-art, distributed datacenter. Successfully achieving this goal will provide significant improvement in performance guarantees and resource utilization in current and future datacenters for a wide range of applications with diverse QoE requirements. Moreover, the proposed approach promotes fundamental analysis in guiding the design of robust complex systems and may give rise to new theories for datacenter and cloud computing in general. The involvement of underrepresented minority and women students, and students with disability in this proposed research through existing course offerings and a REU program will enhance their competitiveness in the future job market.

数据中心服务的最终设计目标是支持以用户为中心的服务，满足个人用户的不同体验质量(QOE)要求，同时允许用户之间优化和公平的计算和网络资源共享。然而，实现这一目标绝非易事。现有的数据中心作业调度和资源调配机制是自下而上的点解决方案，仅解决了相关设计挑战的部分方面，因此无法实现上述目标。他表示，拟议的研究将开发一种涉及整个数据中心系统堆栈的自上而下的整体方法，旨在实现双重目标。它从最高层开始，即用户应用程序级别的QOE或用户实用程序要求。然后，它向下工作，将用户级需求映射到数据中心系统级资源需求，与数据中心计算和网络资源约束一起，为拟议的用户效用感知优化框架设置边界条件。最后，将为一个最先进的分布式数据中心开发一个基于软件的设计和实施所建议的解决方案的示例。成功实现这一目标将显著提高当前和未来数据中心的性能保证和资源利用率，适用于具有不同QOE要求的各种应用。此外，提出的方法促进了基本分析在指导稳健复杂系统设计方面的作用，并可能在总体上为数据中心和云计算带来新的理论。通过现有的课程和REU计划，少数族裔和女学生以及残疾学生参与这项拟议的研究，将增强他们在未来就业市场的竞争力。

项目成果

PACCP: A Price-Aware Congestion Control Protocol for Datacenters

PACCP：数据中心的价格感知拥塞控制协议

• 发表时间: 2020
• 期刊: the 7th The Seventh IEEE/ACM International Conference on Utility and Cloud Computing
• 作者: Sun, Xiaocui;Wang, Zhijun;Wu, Yunxiang;Che, Hao;Jiang, Hong
• 通讯作者: Jiang, Hong

Improving Flash Memory Performance and Reliability for Smartphones With I/O Deduplication

通过 I/O 重复数据删除提高智能手机闪存性能和可靠性

• DOI: 10.1109/tcad.2018.2834395
• 发表时间: 2019-06
• 期刊: IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (CCF A类期刊)
• 作者: Bo Mao;Jindong Zhou;Suzhen Wu;Hong Jiang;Xiao Chen;Weijian Yang
• 通讯作者: Weijian Yang

ForkTail: a black-box fork-join tail latency prediction model for user-facing datacenter workloads

ForkTail：用于面向用户的数据中心工作负载的黑盒分叉连接尾部延迟预测模型

• DOI: 10.1145/3208040.3208058
• 发表时间: 2018
• 期刊: Proceedings of the 27th International Symposium on High-Performance Parallel and Distributed Computing
• 作者: Nguyen, Minh;Alesawi, Sami;Li, Ning;Che, Hao;Jiang, Hong
• 通讯作者: Jiang, Hong

JADE: Tail-Latency-SLO-Aware Job Scheduling for Sensing-as-a-Service

• DOI: 10.1109/ucc48980.2020.00058
• 发表时间: 2020-12
• 期刊: 2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)
• 作者: Stoddard Rosenkrantz;Huiyang Li;Prathyusha Enganti;Zhongwei Li;Lin Sun;Zhijun Wang;Hao Che

DuoModel: Leveraging Reduced Model for Data Reduction and Re-computation on HPC Storage

DuoModel：利用简化模型在 HPC 存储上进行数据缩减和重新计算

• 发表时间: 2018
• 期刊: IEEE letters of the Computer Society
• 作者: Luo, Huizhang;Liu, Qing;Qiao, Zhenbo;Wang, Jinzhen;Wang, Mengxiao;Jiang, Hong
• 通讯作者: Jiang, Hong