English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Though Remote Direct Memory Access (RDMA) promises to reduce datacenternetwork latencies significantly compared to TCP (e.g., 10x), end-to-endcongestion control in the presence of incasts is a challenge. Targeting thefull generality of the congestion problem, previous schemes rely on slow,iterative convergence to the appropriate sending rates (e.g., TIMELY takes 50RTTs). Several papers have shown that even in oversubscribed datacenternetworks most congestion occurs at the receiver. Accordingly, we propose adivide-and-specialize approach, called Dart, which isolates the common case ofreceiver congestion and further subdivides the remaining in-network congestioninto the simpler spatially-localized and the harder spatially-dispersed cases.For receiver congestion, we propose direct apportioning of sending rates (DASR)in which a receiver for n senders directs each sender to cut its rate by afactor of n, converging in only one RTT. For the spatially-localized case, Dartprovides fast (under one RTT) response by adding novel switch hardware forin-order flow deflection (IOFD) because RDMA disallows packet reordering onwhich previous load balancing schemes rely. For the uncommonspatially-dispersed case, Dart falls back to DCQCN. Small-scale testbedmeasurements and at-scale simulations, respectively, show that Dart achieves60% (2.5x) and 79% (4.8x) lower 99th-percentile latency, and similar and 58%higher throughput than InfiniBand, and TIMELY and DCQCN.

Dart: Divide and Specialize for Fast Response to Congestion in RDMA-based Datacenter Networks