V
主页
【ISCA'21】Confidential Serverless Made Efficient with Plug-In Enclaves
发布人
【ISCA'21】Confidential Serverless Made Efficient with Plug-In Enclaves 论文链接:https://ipads.se.sjtu.edu.cn/zh/publications/LiISCA21.pdf 无服务计算近些年来受到工业界和学术界的广泛关注。对于涉及隐私数据的操作,通常采用Enclave对计算过程进行保护,如Intel SGX等。然而,现有的Enclave硬件设计无法很好地满足无服务计算在启动时延等方面的要求,例如,Intel SGX任务的安全加载、远程认证等操作消耗的时间,甚至大于任务本身执行的时间。 本文工作对相关Serverless应用程序进行了性能的量化分析,发现Enclave环境下函数任务的执行时间是普通环境的5倍~423倍。为了解决该问题,该工作基于Intel SGX硬件设计引入了全新抽象——Plug-In Enclave(PIE)。PIE允许以域(Region)为粒度,安全高效地复用预先被验证的Plug-In Enclave,并通过重映射机制实现无数据迁移的原位计算。实验结果表明,PIE设计可以将基于Enclave的无服务计算任务的时延降低94.74%~99.57%,同时将自动伸缩扩容的吞吐提升19倍~179倍。
打开封面
下载高清视频
观看高清视频
视频下载器
【USENIX Security '22】 A Hardware-Software Co-design for Efficient Intra-Enclave
【MICRO'23】Accelerating Extra Dimensional Page Walks for Confidential Computing
【OSDI'21】Penglai: Scalable Enclaves for Modular Applications
【OSDI24】Fast and Scalable In-network Lock Management Using Lock Fission
【NSDI'21】Unifying Timestamp with Transaction Ordering for MVCC with DST
【ATC'23】Bifrost: Analysis and Optimization of Network I/O Tax in Confidential...
【FAST '22】 HTMFS: Strong Consistency Comes for Free with HTM in PMFSes
【ASPLOS'23】BeeHive: Sub-second Elasticity for Web Services with Semi-FaaS Execut
USENIX ATC '20 - UnderBridge
2.5 系统调用
3.3 微内核架构
【HPCA'23】Efficient Distributed Secure Memory with Migratable Merkle Tree
4.2 进程“同时”运行的奥秘:分时复用与上下文切换
3.4 外核Exokernel
4.5_为什么要引入线程?线程有什么特点?
【OSDI'21】Retrofitting High Availability Mechanism to Tame Hybrid Transaction/...
【ATC'22】EPK: Scalable and Efficient Memory Protection Key
【OSDI'22】Microsecond-scale Preemption for Concurrent GPU-accelerated DNN Infe...
【EuroSys21】Bridging the Performance Gap for Copy-based GC atop NVM
操作系统实验讲解:进程与线程
肖文聪:Rubick:Exploiting Job Reconfigurability for Deep Learning Cluster Scheduling
【ATC'22】KRCORE: A Microsecond-scale RDMA Control Plane for Elastic Computing
1.3 macOS的演变:从 POWER 到 X86 再到 ARM
【OSDI'21】Bringing Decentralized Search to Decentralized Services
【APNET'23】Microsecond-scale Datacenter Computing with RDMA: Characterization...
3.1 系统复杂性与操作系统结构
1.1 操作系统的定义
5.2 虚拟内存
5.3 物理内存分配
IPADS OSDI23论文介绍之Flux
3.5_多内核复内核(Multikernel)
5.4 内存管理功能
【FAST '22】MT2: Memory Bandwidth Regulation on Hybrid NVM/DRAM Platforms
1.4 ChCore架构简介
【ATC '23】Bridging the Gap between Relational OLTP and Graph-based OLAP
王则可:Smart Network enhanced AI Systems
【PPoPP'20】No Barrier in the Road: A Comprehensive Study and Optimization of ...
4.3 进程的创建
5.1 内存管理的挑战
1.2 操作系统简介