{"ok":true,"nota":{"id":"521a3e8d0b0c2902","slug":"un-experimento-de-redes-de-almacenamiento-encuentra-un-punto-preciso-d-521a3e","titulo":"Un experimento de redes de almacenamiento encuentra un punto preciso donde el sistema empieza a fallar","resumen":"Investigadores midieron operaciones de transferencia de datos en redes de servidores de alta velocidad. Registraron cuándo y cómo se pierden paquetes y cómo el sistema se recupera. Publicaron los registros crudos para que otros verifiquen las tablas del trabajo.","fuente":"zenodo","fuenteNombre":"Zenodo · Investigación abierta","color":"#1682d4","autor":"Wang, Haoyu, Tran, Bang, Jiang, Peng","url":"https://doi.org/10.5281/zenodo.22844909","imagen":"/hoy/portada/521a3e8d0b0c2902","tema":"tecnologia","subtemas":["redes de datos","almacenamiento","centros de datos","reproducibilidad","pérdida de paquetes"],"entidades":["Zenodo","CloudLab","ICNC 2027"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-19T02:38:11.000Z","indexTs":"2026-10-21T10:34:42.643Z","hash":"4d8bbfd24d5c52ef3a00897e9fac4410fdbbd65a11e487424318940af0158bc4","tsa":{"time":"Oct 21 10:35:48 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"normal","registro":"hallazgo","porQueImporta":"Este trabajo es casi todo infraestructura técnica: sirve para quien opera centros de datos y no cambia nada para el usuario común hoy. Lo relevante es que el resultado depende mucho de la configuración, así que una cifra sirve solo para el equipo donde se midió. Para el ecosistema tecnológico local, es un recordatorio de que el rendimiento de un sistema no se puede copiar de un lugar a otro sin volver a medirlo.","quienFirma":"Zenodo (repositorio de datos asociado a un paper enviado a ICNC 2027)","confianza":"preprint","consulta":null},"sello":{"hash":"4d8bbfd24d5c52ef3a00897e9fac4410fdbbd65a11e487424318940af0158bc4","sello":"cb1ab60416cdf64a76fe6e9d3ba7f88b794d02a7cedd8c24c68fceb9008fbc4a","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6571596,"aleatorio":"2c36e18ae584b00d9a77fc6ee91bac2d1e017f0e506ff9533e9d65bd838a5467","tiempo":"2026-10-21T10:35:00.000Z"},"nist":{"cadena":2,"pulso":1993004,"salida":"23161ED84C8333D8735666F0B82E202695897452E701B068B63EF76A2E59D22F1B819FE2CF27AC06B5894483118E32D36F4246E7C1D82B4789A57D7588B065D3","tiempo":"2026-10-21T10:35:00.000Z"},"noAntes":"2026-10-21T10:35:00.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 21 10:35:48 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Raw measurement logs for \"A Sharp but Configuration-Dependent Loss-Recovery Onset in Closed-Loop RDMA Storage Fabrics\"","descripcion":"<p>Raw per-operation measurement logs for the paper \"A Sharp but Configuration-Dependent Loss-Recovery Onset in Closed-Loop RDMA Storage Fabrics\" (submitted to ICNC 2027). The measurements are NVMe-oF over RoCEv2 on CloudLab c6525-25g nodes (ConnectX-5, 25 GbE), one target and N initiators on a single switch. The archive holds one directory per run: per-operation latency logs, per-run summaries, and NIC hardware-counter snapshots. Every table in the paper is regenerated from these logs by plot_boundary.py in the companion code repository. README-data.md in the archive describes the layout; SHA256SUMS verifies every file. Version 1.1 adds dsscc-icnc-data-msrand.tar.gz: the randomised-order batch-makespan campaign (testbed/ms-rand) behind the paper's batch-completion result, with its own SHA256SUMS-ms-rand. Unpack both archives into the same directory; the original archive is unchanged.</p","pubDate":"2026-10-19T02:38:11Z","url":"https://doi.org/10.5281/zenodo.22844909"},"indexado":"2026-10-21T10:34:42.643Z"}}