MIND: In-Network Memory Management for Disaggregated Data Centers
Seung-seob Lee,Yanpeng Yu,Yupeng Tang,Anurag Khandelwal,Lin Zhong,Abhishek Bhattacharjee +5 more
- pp 488-504
Reads0
Chats0
TLDR
MIND1 as discussed by the authors is an in-network memory management unit for rack-scale disaggregation, which enables transparent resource elasticity while matching the performance of prior memory disaggregation proposals for real-world workloads.Abstract:
Memory disaggregation promises transparent elasticity, high resource utilization and hardware heterogeneity in data centers by physically separating memory and compute into network-attached resource "blades". However, existing designs achieve performance at the cost of resource elasticity, restricting memory sharing to a single compute blade to avoid costly memory coherence traffic over the network. In this work, we show that emerging programmable network switches can enable an efficient shared memory abstraction for disaggregated architectures by placing memory management logic in the network fabric. We find that centralizing memory management in the network permits bandwidth and latency-efficient realization of in-network cache coherence protocols, while programmable switch ASICs support other memory management logic at line-rate. We realize these insights into MIND1, an in-network memory management unit for rack-scale disaggregation. MIND enables transparent resource elasticity while matching the performance of prior memory disaggregation proposals for real-world workloads.read more
Citations
More filters
Proceedings Article
Direct Access, High-Performance Memory Disaggregation with DirectCXL
TL;DR: This paper proposes directly accessible memory disaggregation, D IRECT CXL that straight connects a host processor complex and remote memory resources over CXL’s memory protocol and offers CXL software runtime that allows users to utilize the underlying disaggregated memory resources via sheer load/store instructions.
Journal ArticleDOI
First-generation Memory Disaggregation for Cloud Platforms
Huaicheng Li,Daniel S. Berger,Stanko Novakovic,Lisa R. Hsu,Daniel J. Ernst,Pantea Zardoshti,Monish Shah,Ishwar Agarwal,Mark D. Hill,Marcus Fontoura,Ricardo Bianchini +10 more
TL;DR: This paper proposes a first-generation CXL-based disaggregation system that meets the requirements of cloud providers, and shows that disaggregation can achieve a 9-10% reduction in overall DRAM, which represents hundreds of millions of dollars in cost savings for a large cloud provider.
Proceedings ArticleDOI
Jiffy: elastic far-memory for stateful serverless analytics
TL;DR: Jiffy is presented, an elastic far-memory system for stateful serverless analytics that meets the instantaneous memory demand of a job at seconds timescales, and efficiently multiplexes memory capacity across concurrently running jobs, reducing the overheads of reads and writes to slower persistent storage.
Proceedings ArticleDOI
uBFT: Microsecond-Scale BFT using Disaggregated Memory
Marcos K. Aguilera,Naama Ben-David,Rachid Guerraoui,A. Murat,Athanasios Xygkis,Igor Zablotchi +5 more
TL;DR: In this article , the authors propose uBFT, the first state machine replication (SMR) system to achieve microsecond-scale latency in data centers, while using only 2f+1 replicas to tolerate f Byzantine failures.
Proceedings ArticleDOI
Towards a fully disaggregated and programmable data center
TL;DR: This paper decomposes the overarching problem into four sub-problems and proposes solutions to each of them, and explores the possibility of building a fully disaggregated data center, where every type of resource is disaggregate.
References
More filters
Proceedings Article
The PageRank Citation Ranking : Bringing Order to the Web
TL;DR: This paper describes PageRank, a mathod for rating Web pages objectively and mechanically, effectively measuring the human interest and attention devoted to them, and shows how to efficiently compute PageRank for large numbers of pages.
Proceedings ArticleDOI
What is Twitter, a social network or a news media?
TL;DR: In this paper, the authors have crawled the entire Twittersphere and found a non-power-law follower distribution, a short effective diameter, and low reciprocity, which all mark a deviation from known characteristics of human social networks.
Journal ArticleDOI
The Stanford Dash multiprocessor
Daniel E. Lenoski,James Laudon,Kourosh Gharachorloo,Wolf-Dietrich Weber,Abhinav Gupta,John L. Hennessy,Mark Horowitz,Monica S. Lam +7 more
TL;DR: The directory architecture for shared memory (Dash) as discussed by the authors allows shared data to be cached, significantly reducing the latency of memory accesses and yielding higher processor utilization and higher overall performance, and a distributed directory-based protocol that provides cache coherence without compromising scalability.
Proceedings ArticleDOI
Forwarding metamorphosis: fast programmable match-action processing in hardware for SDN
Pat Bosshart,Glen Gibb,Hun-Seok Kim,George Varghese,Nick McKeown,Martin J. Izzard,Fernando A. Mujica,Mark Horowitz +7 more
TL;DR: The RMT (reconfigurable match tables) model is proposed, a new RISC-inspired pipelined architecture for switching chips, and the essential minimal set of action primitives to specify how headers are processed in hardware are identified.
Proceedings ArticleDOI
The SGI Origin: a ccNUMA highly scalable server
James Laudon,Daniel E. Lenoski +1 more
TL;DR: The motivation for building the Origin 2000 is discussed and the architecture and implementation of the multiprocessor is described, and performance results are presented for the NAS Parallel Benchmarks V2.2 and the SPLASH2 applications.