CAREER: Rebuilding the Virtual Memory Abstraction Across Hardware and Operating Systems

职业：跨硬件和操作系统重建虚拟内存抽象

• 批准号: 2239311
• 负责人: Dimitrios Skarlatos
• 金额: $ 58.78万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-04-01 至 2028-03-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2239311&HistoricalAwards=false
• 关键词: CAREER; Rebuilding; Virtual; Memory; Abstraction

Virtual memory is a cornerstone abstraction of modern computing systems that enables virtualization, programmability, and isolation of memory resources. However, existing virtual memory mechanisms have not been designed for the current era of datacenter computing with ample memory capacity, a plethora of heterogeneous hardware resources, and microarchitectural attacks that are able to bypass current isolation mechanisms. To address these inefficiencies this project will rebuild virtual memory from the ground up. The proposed research agenda will enable high utilization of hardware resources and greatly improve the security of software and hardware systems leading to efficient, sustainable, and secure datacenters. The project includes a comprehensive education plan that is integrated with the proposed research agenda. This includes undergraduate and graduate course offerings, research opportunities, and outreach activities. The research project tackles fundamental research challenges in designing and building a pliable virtual memory abstraction that is scalable, heterogeneous, and secure. The technical approach is based on a virtual memory abstraction that will enable: (i) scalable address translation for terabyte-scale memory systems, (ii) heterogeneous hardware-aware virtual memory support across the stack, and (iii) strong and configurable security for multi-tenant environments in the presence of microarchitectural vulnerabilities.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

虚拟内存是现代计算系统的基石抽象，可实现虚拟化，可编程性和内存资源的隔离。但是，现有的虚拟内存机制尚未针对具有足够的内存容量，大量异质硬件资源和能够绕过电流隔离机制的微体系攻击的当前数据中心计算时代设计。为了解决这些效率低下，该项目将从头开始重建虚拟内存。拟议中的研究议程将能够高利用硬件资源，并大大提高软件和硬件系统的安全性，从而实现高效，可持续和安全的数据中心。该项目包括与拟议的研究议程集成的综合教育计划。这包括本科和研究生课程，研究机会和外展活动。 该研究项目应对设计和建立可扩展，异质和安全的柔韧性虚拟记忆抽象提出的基本研究挑战。技术方法基于一个可启用的虚拟记忆抽象：（i）可伸缩的地址转换，可伸缩的地址尺度记忆系统，（ii）在整个堆栈中的异质性硬件 - 无效的虚拟内存支持，（iii）在多个较大的环境中，在微观统计的情况下均具有较高的环境。使用基金会的智力优点和更广泛的影响评估标准进行评估。

项目成果

Memento: Architectural Support for Ephemeral Memory Management in Serverless Environments

• DOI: 10.1145/3613424.3623795
• 发表时间: 2023-10
• 期刊: 2023 56th IEEE/ACM International Symposium on Microarchitecture (MICRO)
• 作者: Ziqi Wang;Kaiyang Zhao;Pei Li;Andrew Jacob;Michael Kozuch;Todd Mowry;Dimitrios Skarlatos

Contiguitas: The Pursuit of Physical Memory Contiguity in Datacenters

• DOI: 10.1145/3579371.3589079
• 发表时间: 2023-06
• 期刊: Proceedings of the 50th Annual International Symposium on Computer Architecture
• 作者: Kaiyang Zhao;Kaiwen Xue;Ziqi Wang-;Dan Schatzberg;Leon Yang;Antonis Manousis;Johannes Weiner;Rik Van Riel;Bikash Sharma;Chunqiang Tang;Dimitrios Skarlatos