REPS: Recycling Entropies for Packet Spraying to Adaptively Explore Paths and Mitigate Failures
Authors:
Tommaso Bonato,
Abdul Kabbani,
Ahmad Ghalayini,
Mohammad Dohadwala,
Michael Papamichael,
Daniele De Sensi,
Torsten Hoefler
Abstract:
Most existing datacenter transport protocols rely on in-order packet delivery, a design choice rooted in legacy systems and simplicity. However, advancements in technology, such as RDMA, have made it feasible to relax this requirement, allowing for more effective use of modern datacenter topologies like FatTree and Dragonfly. The rise of AI/ML workloads underscores the necessity for enhanced link…
▽ More
Most existing datacenter transport protocols rely on in-order packet delivery, a design choice rooted in legacy systems and simplicity. However, advancements in technology, such as RDMA, have made it feasible to relax this requirement, allowing for more effective use of modern datacenter topologies like FatTree and Dragonfly. The rise of AI/ML workloads underscores the necessity for enhanced link utilization, a challenge for single-path load balancers due to issues like ECMP collisions.
In this paper, we introduce REPS, a novel per-packet traffic load-balancing algorithm that integrates seamlessly with existing congestion control mechanisms. REPS reroutes packets around congested hotspots and unreliable or failing links with remarkable simplicity and minimal state requirements.
Our evaluation demonstrates that REPS significantly outperforms traditional packet spraying and other state-of-the-art solutions in datacenter networks, offering substantial improvements in performance and link utilization.
△ Less
Submitted 31 July, 2024;
originally announced July 2024.
SMaRTT-REPS: Sender-based Marked Rapidly-adapting Trimmed & Timed Transport with Recycled Entropies
Authors:
Tommaso Bonato,
Abdul Kabbani,
Daniele De Sensi,
Rong Pan,
Yanfang Le,
Costin Raiciu,
Mark Handley,
Timo Schneider,
Nils Blach,
Ahmad Ghalayini,
Daniel Alves,
Michael Papamichael,
Adrian Caulfield,
Torsten Hoefler
Abstract:
With the rapid growth of machine learning (ML) workloads in datacenters, existing congestion control (CC) algorithms fail to deliver the required performance at scale. ML traffic is bursty and bulk-synchronous and thus requires quick reaction and strong fairness. We show that existing CC algorithms that use delay as a main signal react too slowly and are not always fair. We design SMaRTT, a simple…
▽ More
With the rapid growth of machine learning (ML) workloads in datacenters, existing congestion control (CC) algorithms fail to deliver the required performance at scale. ML traffic is bursty and bulk-synchronous and thus requires quick reaction and strong fairness. We show that existing CC algorithms that use delay as a main signal react too slowly and are not always fair. We design SMaRTT, a simple sender-based CC algorithm that combines delay, ECN, and optional packet trimming for fast and precise window adjustments. At the core of SMaRTT lies the novel QuickAdapt algorithm that accurately estimates the bandwidth at the receiver. We show how to combine SMaRTT with a new per-packet traffic load-balancing algorithm called REPS to effectively reroute packets around congested hotspots as well as flaky or failing links. Our evaluation shows that SMaRTT alone outperforms EQDS, Swift, BBR, and MPRDMA by up to 50% on modern datacenter networks.
△ Less
Submitted 27 April, 2024; v1 submitted 2 April, 2024;
originally announced April 2024.
