24 lines
35 KiB
HTML
24 lines
35 KiB
HTML
<!doctype html><html lang=en dir=auto data-theme=auto><head><meta charset=utf-8><meta http-equiv=X-UA-Compatible content="IE=edge"><meta name=viewport content="width=device-width,initial-scale=1,shrink-to-fit=no"><meta name=robots content="index, follow"><title>NSDI 2025 Digest | Publish Assistant</title><meta name=keywords content="networking,distributed-systems,cloud,programmable-data-planes,network-verification,datacenter,transport,ml-systems,5g"><meta name=description content="13 papers selected.
|
||
|
||
PRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters
|
||
Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu et al.
|
||
TL;DR — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.
|
||
Why notable — RED has been a cornerstone of congestion control for decades; PRED’s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches."><meta name=author content="Publish Assistant"><link rel=canonical href=https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/nsdi-2025/><link crossorigin=anonymous href=/vincent/publish-assistant/assets/css/stylesheet.d72f07832e13c592b3edba91680bfe70f01daac396179bcace0ac36e8e0494c6.css integrity="sha256-1y8Hgy4TxZKz7bqRaAv+cPAdqsOWF5vKzgrDbo4ElMY=" rel="preload stylesheet" as=style><link rel=icon href=https://pub.sqrt.fr/vincent/publish-assistant/favicon.ico><link rel=icon type=image/png sizes=16x16 href=https://pub.sqrt.fr/vincent/publish-assistant/favicon-16x16.png><link rel=icon type=image/png sizes=32x32 href=https://pub.sqrt.fr/vincent/publish-assistant/favicon-32x32.png><link rel=apple-touch-icon href=https://pub.sqrt.fr/vincent/publish-assistant/apple-touch-icon.png><link rel=mask-icon href=https://pub.sqrt.fr/vincent/publish-assistant/safari-pinned-tab.svg><meta name=theme-color content="#2e2e33"><meta name=msapplication-TileColor content="#2e2e33"><link rel=alternate hreflang=en href=https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/nsdi-2025/><noscript><style>#theme-toggle,.top-link{display:none}</style><style>@media(prefers-color-scheme:dark){:root{--theme:rgb(29, 30, 32);--entry:rgb(46, 46, 51);--primary:rgb(218, 218, 219);--secondary:rgb(155, 156, 157);--tertiary:rgb(65, 66, 68);--content:rgb(196, 196, 197);--code-block-bg:rgb(46, 46, 51);--code-bg:rgb(55, 56, 62);--border:rgb(51, 51, 51);color-scheme:dark}.list{background:var(--theme)}.toc{background:var(--entry)}}</style></noscript><script>localStorage.getItem("pref-theme")==="dark"?document.querySelector("html").dataset.theme="dark":localStorage.getItem("pref-theme")==="light"?document.querySelector("html").dataset.theme="light":window.matchMedia("(prefers-color-scheme: dark)").matches?document.querySelector("html").dataset.theme="dark":document.querySelector("html").dataset.theme="light"</script><meta property="og:url" content="https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/nsdi-2025/"><meta property="og:site_name" content="Publish Assistant"><meta property="og:title" content="NSDI 2025 Digest"><meta property="og:description" content="13 papers selected.
|
||
PRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu et al.
|
||
TL;DR — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.
|
||
Why notable — RED has been a cornerstone of congestion control for decades; PRED’s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches."><meta property="og:locale" content="en_us"><meta property="og:type" content="article"><meta property="article:section" content="cloud-edge"><meta property="article:published_time" content="2025-04-28T00:00:00+00:00"><meta property="article:modified_time" content="2025-04-28T00:00:00+00:00"><meta property="article:tag" content="Networking"><meta property="article:tag" content="Distributed-Systems"><meta property="article:tag" content="Cloud"><meta property="article:tag" content="Programmable-Data-Planes"><meta property="article:tag" content="Network-Verification"><meta property="article:tag" content="Datacenter"><meta name=twitter:card content="summary"><meta name=twitter:title content="NSDI 2025 Digest"><meta name=twitter:description content="13 papers selected.
|
||
PRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu et al.
|
||
TL;DR — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.
|
||
Why notable — RED has been a cornerstone of congestion control for decades; PRED’s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches."><script type=application/ld+json>{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Edge and Cloud Systems","item":"https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/"},{"@type":"ListItem","position":2,"name":"Digests","item":"https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/"},{"@type":"ListItem","position":3,"name":"NSDI 2025 Digest","item":"https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/nsdi-2025/"}]}</script><script type=application/ld+json>{"@context":"https://schema.org","@type":"BlogPosting","headline":"NSDI 2025 Digest","name":"NSDI 2025 Digest","description":"13 papers selected.\nPRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu et al.\nTL;DR — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.\nWhy notable — RED has been a cornerstone of congestion control for decades; PRED\u0026rsquo;s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches.\n","keywords":["networking","distributed-systems","cloud","programmable-data-planes","network-verification","datacenter","transport","ml-systems","5g"],"articleBody":"13 papers selected.\nPRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu et al.\nTL;DR — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.\nWhy notable — RED has been a cornerstone of congestion control for decades; PRED’s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches.\n→ Read paper Rajomon: Decentralized and Coordinated Overload Control for Latency-Sensitive Microservices Jiali Xing, Akis Giannoukos, Paul Loh, Shuyue Wang et al.\nTL;DR — Rajomon introduces a token-based, decentralized overload control mechanism that coordinates admission across microservice call graphs without a central bottleneck.\nWhy notable — Microservice overload propagation is a persistent pain point in production clouds; Rajomon’s approach of spreading load-shedding decisions across the call graph while retaining global coherence is a practical and principled contribution. The system is evaluated on realistic cloud benchmarks and shows clear SLO improvement over prior centralized and uncoordinated schemes.\n→ Read paper Unlocking ECMP Programmability for Precise Traffic Control Yadong Liu, Yunming Xiao, Xuan Zhang, Weizhen Dang et al.\nTL;DR — This work exposes fine-grained ECMP programming interfaces that allow operators to steer individual flows through specific paths in multipath datacenter fabrics with near-zero overhead.\nWhy notable — ECMP’s coarse hashing has been a limiting factor in datacenter traffic engineering for years; this paper shows that commodity switch primitives can be composed to achieve precise per-flow placement, delivering measurable improvements in load balance without topology changes.\n→ Read paper eTran: Extensible Kernel Transport with eBPF Zhongjie Chen, Qingkai Meng 0001, ChonLam Lao, Yifan Liu et al.\nTL;DR — eTran uses eBPF to let applications plug in custom transport logic—including RDMA-like zero-copy paths—directly into the Linux kernel data path without modifying kernel source.\nWhy notable — The ability to safely extend kernel networking with eBPF has been widely anticipated; eTran demonstrates that full transport protocols (not just classifiers) can be realised this way, achieving performance competitive with kernel-bypass while retaining OS isolation and deployability.\n→ Read paper White-Boxing RDMA with Packet-Granular Software Control Chenxingyu Zhao, Jaehong Min, Ming Liu 0027, Arvind Krishnamurthy\nTL;DR — This paper decomposes RDMA semantics and re-implements them in software at packet granularity, exposing hooks that allow transport policy (retransmission, congestion, multipath) to be modified without hardware changes.\nWhy notable — RDMA’s black-box nature has long frustrated datacenter operators trying to deploy custom congestion control or network telemetry; white-boxing RDMA at packet granularity is a principled answer that simultaneously improves debuggability and extensibility, with demonstrated line-rate performance.\n→ Read paper NDD: A Decision Diagram for Network Verification Zechun Li, Peng Zhang 0011, Yichi Zhang, Hongkun Yang\nTL;DR — NDD is a new symbolic data structure that compactly represents the forwarding behavior of large networks, enabling network verification queries orders of magnitude faster than BDD-based predecessors.\nWhy notable — Network verification tools like Batfish and ARC rely on set representations whose size can explode on real networks; NDD’s topology-aware decomposition dramatically reduces verification time for common queries (reachability, loop detection) and is likely to be adopted as a backend in production verification pipelines.\n→ Read paper VEP: A Two-stage Verification Toolchain for Full eBPF Programmability Xiwei Wu, Yueyang Feng, Tianyi Huang, Xiaoyang Lu et al.\nTL;DR — VEP combines abstract interpretation with deductive verification to prove safety properties of eBPF programs that the Linux kernel verifier rejects, enabling a much wider class of programs to run safely in the kernel.\nWhy notable — The Linux eBPF verifier’s conservatism blocks many useful programs; VEP’s two-stage approach (lightweight in-kernel check plus offline proof) expands the programmability frontier without weakening safety guarantees, directly affecting every operator who writes eBPF for networking or observability.\n→ Read paper MTP: Transport for In-Network Computing Tao Ji, Rohan Vardekar, Balajee Vamanan, Brent E. Stephens et al.\nTL;DR — MTP is a transport protocol co-designed with programmable switches that allows in-network compute operations (aggregation, filtering) to be expressed as first-class transport primitives with reliability and flow-control guarantees.\nWhy notable — Existing transports treat the network as a dumb pipe and retrofit in-network compute as an afterthought; MTP shows that rethinking transport abstractions around programmable hardware can yield substantial throughput gains for distributed ML and key-value workloads.\n→ Read paper State-Compute Replication: Parallelizing High-Speed Stateful Packet Processing Qiongwen Xu, Sebastiano Miano, Xiangyu Gao, Tao Wang 0088 et al.\nTL;DR — This paper introduces a replication model for stateful packet processing that partitions state across multiple pipeline replicas and reconciles updates at line rate, breaking the single-pipeline bottleneck of P4 programs.\nWhy notable — Stateful P4 programs that require per-flow counters or heavy-hitter detection are bottlenecked by single-instance state; the state-compute replication model is a general technique applicable across telemetry, load balancing, and security use cases, with a prototype running at 100 Gbps.\n→ Read paper Efficient Direct-Connect Topologies for Collective Communications Liangyu Zhao, Siddharth Pal, Tapan Chugh, Weiyang Wang et al.\nTL;DR — This paper derives a family of direct-connect network topologies mathematically optimized for the all-reduce and all-to-all collectives used in large-scale ML training, achieving better bandwidth and lower diameter than fat-tree alternatives at the same port count.\nWhy notable — As AI training clusters scale to thousands of GPUs, interconnect topology becomes a first-order performance determinant; the paper’s analytical framework for topology synthesis fills a long-standing gap between graph theory and practical cluster networking, and the proposed topologies outperform fat-trees on real collective benchmarks.\n→ Read paper SimAI: Unifying Architecture Design and Performance Tuning for Large-Scale Large Language Model Training with Scalability and Precision Xizheng Wang, Qingxu Li, Yichi Xu, Gang Lu et al.\nTL;DR — SimAI is a high-fidelity simulator that jointly models compute, network, and storage interactions in LLM training clusters, enabling architects to evaluate topology and parallelism strategy choices without running full cluster experiments.\nWhy notable — Designing training infrastructure for frontier LLMs is prohibitively expensive to explore empirically; SimAI’s validated accuracy against production Alibaba clusters makes it a practical tool for the community and provides rare quantitative insight into how network architecture choices affect end-to-end training throughput.\n→ Read paper Minder: Faulty Machine Detection for Large-scale Distributed Model Training Yangtao Deng, Xiang Shi, Zhuo Jiang, Xingjian Zhang 0009 et al.\nTL;DR — Minder continuously monitors collective-communication timing patterns during LLM training to pinpoint faulty nodes—including subtle slow or flapping hardware—without requiring dedicated diagnostic jobs.\nWhy notable — GPU cluster faults that cause training slowdowns rather than outright crashes are notoriously hard to attribute; Minder’s passive monitoring approach catches a wide class of hardware anomalies (including network stragglers) with sub-minute detection latency, and its deployment at scale in Alibaba’s training infrastructure gives the results strong operational credibility.\n→ Read paper Efficient Multi-WAN Transport for 5G with OTTER Mary Hogan, Gerry Wan, Yiming Qiu, Sharad Agarwal et al.\nTL;DR — OTTER is a multi-path transport system for 5G user-plane traffic that dynamically bonds heterogeneous WAN links (e.g., terrestrial + satellite) while hiding link asymmetry and jitter from applications.\nWhy notable — 5G deployments increasingly rely on diverse backhaul paths with wildly different characteristics; OTTER’s scheduler, validated on a Microsoft production 5G deployment, demonstrates that significant throughput and reliability gains are achievable through principled multi-WAN design, informing both operator practice and future transport standards.\n→ Read paper ","wordCount":"1238","inLanguage":"en","datePublished":"2025-04-28T00:00:00Z","dateModified":"2025-04-28T00:00:00Z","author":{"@type":"Person","name":"Publish Assistant"},"mainEntityOfPage":{"@type":"WebPage","@id":"https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/nsdi-2025/"},"publisher":{"@type":"Organization","name":"Publish Assistant","logo":{"@type":"ImageObject","url":"https://pub.sqrt.fr/vincent/publish-assistant/favicon.ico"}}}</script></head><body id=top><header class=header><nav class=header-nav><div class=logo><a href=https://pub.sqrt.fr/vincent/publish-assistant/ accesskey=h title="Publish Assistant (Alt + H)">Publish Assistant</a>
|
||
<span class=logo-sep>/</span>
|
||
<a class=logo-topic href=/vincent/publish-assistant/cloud-edge/ title="Edge and Cloud Systems">Edge and Cloud Systems</a><div class=logo-switches><button id=theme-toggle class=theme-toggle accesskey=t title="(Alt + T)" aria-label="Toggle theme">
|
||
<svg class="moon" width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M21 12.79A9 9 0 1111.21 3 7 7 0 0021 12.79z"/></svg>
|
||
<svg class="sun" width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><circle cx="12" cy="12" r="5"/><line x1="12" y1="1" x2="12" y2="3"/><line x1="12" y1="21" x2="12" y2="23"/><line x1="4.22" y1="4.22" x2="5.64" y2="5.64"/><line x1="18.36" y1="18.36" x2="19.78" y2="19.78"/><line x1="1" y1="12" x2="3" y2="12"/><line x1="21" y1="12" x2="23" y2="12"/><line x1="4.22" y1="19.78" x2="5.64" y2="18.36"/><line x1="18.36" y1="5.64" x2="19.78" y2="4.22"/></svg></button></div></div><ul id=menu class=menu><li><a href=/vincent/publish-assistant/cloud-edge/venues/ title=Venues><span>Venues</span></a></li><li><a href=/vincent/publish-assistant/cloud-edge/calendar/ title=Calendar><span>Calendar</span></a></li><li><a href=/vincent/publish-assistant/cloud-edge/digests/ title=Digests><span class=active>Digests</span></a></li></ul></nav></header><main class=main><article class=post-single><header class=post-header><nav class=breadcrumbs role=navigation aria-label=Breadcrumb><a href=/vincent/publish-assistant/cloud-edge/digests/>Digests</a>
|
||
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="feather feather-chevron-right"><polyline points="9 18 15 12 9 6"/></svg></nav><h1 class="post-title entry-hint-parent">NSDI 2025 Digest</h1><div class=post-meta><span title='2025-04-28 00:00:00 +0000 UTC'>April 28, 2025</span> · <span>Publish Assistant</span></div></header><div class="post-content md-content"><p>13 papers selected.</p><hr><h3 id=pred-performance-oriented-random-early-detection-for-consistently-stable-performance-in-datacenters>PRED: Performance-oriented Random Early Detection for Consistently Stable Performance in Datacenters<a hidden class=anchor aria-hidden=true href=#pred-performance-oriented-random-early-detection-for-consistently-stable-performance-in-datacenters>#</a></h3><p><em>Xinle Du, Tong Li 0014, Guangmeng Zhou, Zhuotao Liu <em>et al.</em></em></p><p><strong>TL;DR</strong> — PRED redesigns AQM by making drop probability a direct function of per-flow performance targets rather than queue length, eliminating the instability of classic RED in modern datacenter workloads.</p><p><strong>Why notable</strong> — RED has been a cornerstone of congestion control for decades; PRED’s performance-centric reformulation challenges a long-held design axiom and demonstrates significantly lower tail latency at scale. It opens the door to intent-driven AQM as a first-class primitive in datacenter switches.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/du>→ Read paper</a></p><hr><h3 id=rajomon-decentralized-and-coordinated-overload-control-for-latency-sensitive-microservices>Rajomon: Decentralized and Coordinated Overload Control for Latency-Sensitive Microservices<a hidden class=anchor aria-hidden=true href=#rajomon-decentralized-and-coordinated-overload-control-for-latency-sensitive-microservices>#</a></h3><p><em>Jiali Xing, Akis Giannoukos, Paul Loh, Shuyue Wang <em>et al.</em></em></p><p><strong>TL;DR</strong> — Rajomon introduces a token-based, decentralized overload control mechanism that coordinates admission across microservice call graphs without a central bottleneck.</p><p><strong>Why notable</strong> — Microservice overload propagation is a persistent pain point in production clouds; Rajomon’s approach of spreading load-shedding decisions across the call graph while retaining global coherence is a practical and principled contribution. The system is evaluated on realistic cloud benchmarks and shows clear SLO improvement over prior centralized and uncoordinated schemes.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/xing>→ Read paper</a></p><hr><h3 id=unlocking-ecmp-programmability-for-precise-traffic-control>Unlocking ECMP Programmability for Precise Traffic Control<a hidden class=anchor aria-hidden=true href=#unlocking-ecmp-programmability-for-precise-traffic-control>#</a></h3><p><em>Yadong Liu, Yunming Xiao, Xuan Zhang, Weizhen Dang <em>et al.</em></em></p><p><strong>TL;DR</strong> — This work exposes fine-grained ECMP programming interfaces that allow operators to steer individual flows through specific paths in multipath datacenter fabrics with near-zero overhead.</p><p><strong>Why notable</strong> — ECMP’s coarse hashing has been a limiting factor in datacenter traffic engineering for years; this paper shows that commodity switch primitives can be composed to achieve precise per-flow placement, delivering measurable improvements in load balance without topology changes.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/liu-yadong>→ Read paper</a></p><hr><h3 id=etran-extensible-kernel-transport-with-ebpf>eTran: Extensible Kernel Transport with eBPF<a hidden class=anchor aria-hidden=true href=#etran-extensible-kernel-transport-with-ebpf>#</a></h3><p><em>Zhongjie Chen, Qingkai Meng 0001, ChonLam Lao, Yifan Liu <em>et al.</em></em></p><p><strong>TL;DR</strong> — eTran uses eBPF to let applications plug in custom transport logic—including RDMA-like zero-copy paths—directly into the Linux kernel data path without modifying kernel source.</p><p><strong>Why notable</strong> — The ability to safely extend kernel networking with eBPF has been widely anticipated; eTran demonstrates that full transport protocols (not just classifiers) can be realised this way, achieving performance competitive with kernel-bypass while retaining OS isolation and deployability.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/chen-zhongjie>→ Read paper</a></p><hr><h3 id=white-boxing-rdma-with-packet-granular-software-control>White-Boxing RDMA with Packet-Granular Software Control<a hidden class=anchor aria-hidden=true href=#white-boxing-rdma-with-packet-granular-software-control>#</a></h3><p><em>Chenxingyu Zhao, Jaehong Min, Ming Liu 0027, Arvind Krishnamurthy</em></p><p><strong>TL;DR</strong> — This paper decomposes RDMA semantics and re-implements them in software at packet granularity, exposing hooks that allow transport policy (retransmission, congestion, multipath) to be modified without hardware changes.</p><p><strong>Why notable</strong> — RDMA’s black-box nature has long frustrated datacenter operators trying to deploy custom congestion control or network telemetry; white-boxing RDMA at packet granularity is a principled answer that simultaneously improves debuggability and extensibility, with demonstrated line-rate performance.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/zhao-chenxingyu>→ Read paper</a></p><hr><h3 id=ndd-a-decision-diagram-for-network-verification>NDD: A Decision Diagram for Network Verification<a hidden class=anchor aria-hidden=true href=#ndd-a-decision-diagram-for-network-verification>#</a></h3><p><em>Zechun Li, Peng Zhang 0011, Yichi Zhang, Hongkun Yang</em></p><p><strong>TL;DR</strong> — NDD is a new symbolic data structure that compactly represents the forwarding behavior of large networks, enabling network verification queries orders of magnitude faster than BDD-based predecessors.</p><p><strong>Why notable</strong> — Network verification tools like Batfish and ARC rely on set representations whose size can explode on real networks; NDD’s topology-aware decomposition dramatically reduces verification time for common queries (reachability, loop detection) and is likely to be adopted as a backend in production verification pipelines.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/li-zechun>→ Read paper</a></p><hr><h3 id=vep-a-two-stage-verification-toolchain-for-full-ebpf-programmability>VEP: A Two-stage Verification Toolchain for Full eBPF Programmability<a hidden class=anchor aria-hidden=true href=#vep-a-two-stage-verification-toolchain-for-full-ebpf-programmability>#</a></h3><p><em>Xiwei Wu, Yueyang Feng, Tianyi Huang, Xiaoyang Lu <em>et al.</em></em></p><p><strong>TL;DR</strong> — VEP combines abstract interpretation with deductive verification to prove safety properties of eBPF programs that the Linux kernel verifier rejects, enabling a much wider class of programs to run safely in the kernel.</p><p><strong>Why notable</strong> — The Linux eBPF verifier’s conservatism blocks many useful programs; VEP’s two-stage approach (lightweight in-kernel check plus offline proof) expands the programmability frontier without weakening safety guarantees, directly affecting every operator who writes eBPF for networking or observability.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/wu-xiwei>→ Read paper</a></p><hr><h3 id=mtp-transport-for-in-network-computing>MTP: Transport for In-Network Computing<a hidden class=anchor aria-hidden=true href=#mtp-transport-for-in-network-computing>#</a></h3><p><em>Tao Ji, Rohan Vardekar, Balajee Vamanan, Brent E. Stephens <em>et al.</em></em></p><p><strong>TL;DR</strong> — MTP is a transport protocol co-designed with programmable switches that allows in-network compute operations (aggregation, filtering) to be expressed as first-class transport primitives with reliability and flow-control guarantees.</p><p><strong>Why notable</strong> — Existing transports treat the network as a dumb pipe and retrofit in-network compute as an afterthought; MTP shows that rethinking transport abstractions around programmable hardware can yield substantial throughput gains for distributed ML and key-value workloads.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/ji>→ Read paper</a></p><hr><h3 id=state-compute-replication-parallelizing-high-speed-stateful-packet-processing>State-Compute Replication: Parallelizing High-Speed Stateful Packet Processing<a hidden class=anchor aria-hidden=true href=#state-compute-replication-parallelizing-high-speed-stateful-packet-processing>#</a></h3><p><em>Qiongwen Xu, Sebastiano Miano, Xiangyu Gao, Tao Wang 0088 <em>et al.</em></em></p><p><strong>TL;DR</strong> — This paper introduces a replication model for stateful packet processing that partitions state across multiple pipeline replicas and reconciles updates at line rate, breaking the single-pipeline bottleneck of P4 programs.</p><p><strong>Why notable</strong> — Stateful P4 programs that require per-flow counters or heavy-hitter detection are bottlenecked by single-instance state; the state-compute replication model is a general technique applicable across telemetry, load balancing, and security use cases, with a prototype running at 100 Gbps.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/xu-qiongwen>→ Read paper</a></p><hr><h3 id=efficient-direct-connect-topologies-for-collective-communications>Efficient Direct-Connect Topologies for Collective Communications<a hidden class=anchor aria-hidden=true href=#efficient-direct-connect-topologies-for-collective-communications>#</a></h3><p><em>Liangyu Zhao, Siddharth Pal, Tapan Chugh, Weiyang Wang <em>et al.</em></em></p><p><strong>TL;DR</strong> — This paper derives a family of direct-connect network topologies mathematically optimized for the all-reduce and all-to-all collectives used in large-scale ML training, achieving better bandwidth and lower diameter than fat-tree alternatives at the same port count.</p><p><strong>Why notable</strong> — As AI training clusters scale to thousands of GPUs, interconnect topology becomes a first-order performance determinant; the paper’s analytical framework for topology synthesis fills a long-standing gap between graph theory and practical cluster networking, and the proposed topologies outperform fat-trees on real collective benchmarks.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/zhao-liangyu>→ Read paper</a></p><hr><h3 id=simai-unifying-architecture-design-and-performance-tuning-for-large-scale-large-language-model-training-with-scalability-and-precision>SimAI: Unifying Architecture Design and Performance Tuning for Large-Scale Large Language Model Training with Scalability and Precision<a hidden class=anchor aria-hidden=true href=#simai-unifying-architecture-design-and-performance-tuning-for-large-scale-large-language-model-training-with-scalability-and-precision>#</a></h3><p><em>Xizheng Wang, Qingxu Li, Yichi Xu, Gang Lu <em>et al.</em></em></p><p><strong>TL;DR</strong> — SimAI is a high-fidelity simulator that jointly models compute, network, and storage interactions in LLM training clusters, enabling architects to evaluate topology and parallelism strategy choices without running full cluster experiments.</p><p><strong>Why notable</strong> — Designing training infrastructure for frontier LLMs is prohibitively expensive to explore empirically; SimAI’s validated accuracy against production Alibaba clusters makes it a practical tool for the community and provides rare quantitative insight into how network architecture choices affect end-to-end training throughput.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/wang-xizheng-simai>→ Read paper</a></p><hr><h3 id=minder-faulty-machine-detection-for-large-scale-distributed-model-training>Minder: Faulty Machine Detection for Large-scale Distributed Model Training<a hidden class=anchor aria-hidden=true href=#minder-faulty-machine-detection-for-large-scale-distributed-model-training>#</a></h3><p><em>Yangtao Deng, Xiang Shi, Zhuo Jiang, Xingjian Zhang 0009 <em>et al.</em></em></p><p><strong>TL;DR</strong> — Minder continuously monitors collective-communication timing patterns during LLM training to pinpoint faulty nodes—including subtle slow or flapping hardware—without requiring dedicated diagnostic jobs.</p><p><strong>Why notable</strong> — GPU cluster faults that cause training slowdowns rather than outright crashes are notoriously hard to attribute; Minder’s passive monitoring approach catches a wide class of hardware anomalies (including network stragglers) with sub-minute detection latency, and its deployment at scale in Alibaba’s training infrastructure gives the results strong operational credibility.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/deng>→ Read paper</a></p><hr><h3 id=efficient-multi-wan-transport-for-5g-with-otter>Efficient Multi-WAN Transport for 5G with OTTER<a hidden class=anchor aria-hidden=true href=#efficient-multi-wan-transport-for-5g-with-otter>#</a></h3><p><em>Mary Hogan, Gerry Wan, Yiming Qiu, Sharad Agarwal <em>et al.</em></em></p><p><strong>TL;DR</strong> — OTTER is a multi-path transport system for 5G user-plane traffic that dynamically bonds heterogeneous WAN links (e.g., terrestrial + satellite) while hiding link asymmetry and jitter from applications.</p><p><strong>Why notable</strong> — 5G deployments increasingly rely on diverse backhaul paths with wildly different characteristics; OTTER’s scheduler, validated on a Microsoft production 5G deployment, demonstrates that significant throughput and reliability gains are achievable through principled multi-WAN design, informing both operator practice and future transport standards.</p><p><a href=https://www.usenix.org/conference/nsdi25/presentation/hogan>→ Read paper</a></p></div><footer class=post-footer><ul class=post-tags><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/networking/>Networking</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/distributed-systems/>Distributed-Systems</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/cloud/>Cloud</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/programmable-data-planes/>Programmable-Data-Planes</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/network-verification/>Network-Verification</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/datacenter/>Datacenter</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/transport/>Transport</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/ml-systems/>Ml-Systems</a></li><li><a href=https://pub.sqrt.fr/vincent/publish-assistant/tags/5g/>5g</a></li></ul><nav class=paginav><a class=prev href=https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/ipdps-2025/><span class=title>« Prev</span>
|
||
<span>IPDPS 2025 Digest</span>
|
||
</a><a class=next href=https://pub.sqrt.fr/vincent/publish-assistant/cloud-edge/digests/eurosys-2025/><span class=title>Next »</span>
|
||
<span>EuroSys 2025 Digest</span></a></nav></footer></article></main><footer class=footer><span>© 2026 <a href=https://pub.sqrt.fr/vincent/publish-assistant/>Publish Assistant</a></span> ·
|
||
<span>Powered by
|
||
<a href="https://gohugo.io/?utm_source=papermod" rel=noopener target=_blank>Hugo</a> &
|
||
<a href=https://github.com/adityatelange/hugo-PaperMod/ rel=noopener target=_blank>PaperMod</a></span></footer><a href=#top id=top-link class="top-link hidden" aria-label="go to top" title="Go to Top (Alt + G)" accesskey=g><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="feather feather-chevrons-up"><polyline points="17 11 12 6 7 11"/><polyline points="17 18 12 13 7 18"/></svg>
|
||
</a><script>let menu=document.getElementById("menu");if(menu){const e=localStorage.getItem("menu-scroll-position");e&&(menu.scrollLeft=parseInt(e,10)),menu.onscroll=function(){localStorage.setItem("menu-scroll-position",menu.scrollLeft)}}document.querySelectorAll('a[href^="#"]').forEach(e=>{e.addEventListener("click",function(e){e.preventDefault();var t=this.getAttribute("href").substr(1);window.matchMedia("(prefers-reduced-motion: reduce)").matches?document.querySelector(`[id='${decodeURIComponent(t)}']`).scrollIntoView():document.querySelector(`[id='${decodeURIComponent(t)}']`).scrollIntoView({behavior:"smooth"}),t==="top"?history.replaceState(null,null," "):history.pushState(null,null,`#${t}`)})})</script><script>var toplink=document.getElementById("top-link");window.onscroll=function(){const e=window.innerHeight;document.body.scrollTop>e||document.documentElement.scrollTop>e?toplink.classList.remove("hidden"):toplink.classList.add("hidden")}</script><script>document.getElementById("theme-toggle").addEventListener("click",()=>{const e=document.querySelector("html");e.dataset.theme==="dark"?(e.dataset.theme="light",localStorage.setItem("pref-theme","light")):(e.dataset.theme="dark",localStorage.setItem("pref-theme","dark"))})</script></body></html> |