SDR-RDMA: S oftware- D efined R eliability Architecture for Planetary Scale RDMA Communication
Mikhail Khalilov, Siyuan Shen, Marcin Chrapek, Tiancheng Chen, Kenji Nakano, Peter-Jan Gootzen, Salvatore Di Girolamo, Rami Nudelman, Gil Bloch, Sreevatsa Anantharamu, Mahmoud Elhaddad, Jithin Jose, Abdul Kabbani, Scott Moe, Konstantin Taranov, Zhuolong Yu, Jie Zhang, Nicola Mazzoletti, Torsten Hoef · 2025
Read SDR-RDMA: S oftware- D efined R eliability Architecture for Planetary Scale RDMA Communication by Mikhail Khalilov, Siyuan Shen, Marcin Chrapek, Tiancheng Chen, Kenji Nakano, Peter-Jan Gootzen, Salvatore Di Girolamo, Rami Nudelman, Gil Bloch, Sreevatsa Anantharamu, Mahmoud Elhaddad, Jithin Jose, Abdul Kabbani, Scott Moe, Konstantin Taranov, Zhuolong Yu, Jie Zhang, Nicola Mazzoletti, Torsten Hoef (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-05366.txt.
Description
Mikhail Khalilov and co-authors redesign this 2025 reliability layer for the fast network links that train AI across data centers. RDMA moves data between machines with minimal fuss, but links spanning continents drop packets and lag by milliseconds. The standard fix, resending lost pieces in order, wastes the link under those conditions. Alternatives like erasure coding cannot run on today's hardware semantics.
Their software-defined stack adds a receive-buffer bitmap to ordinary RDMA, letting software see partial arrivals and build its own recovery scheme. A dedicated accelerator chip runs the backend at full line rate, preserving zero-copy speed. The design opens cross-continent training to smarter reliability strategies without new network cards.
Catalog record
| CREATOR | Mikhail Khalilov, Siyuan Shen, Marcin Chrapek, Tiancheng Chen, Kenji Nakano, Peter-Jan Gootzen, Salvatore Di Girolamo, Rami Nudelman, Gil Bloch, Sreevatsa Anantharamu, Mahmoud Elhaddad, Jithin Jose, Abdul Kabbani, Scott Moe, Konstantin Taranov, Zhuolong Yu, Jie Zhang, Nicola Mazzoletti, Torsten Hoef |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-05366