P4-Days: Programmable Data-Plane Execution in a Multicore Discrete-Event Network Simulator

Packet-level simulation of P4-programmable networks is expensive because a software P4 pipeline must be interpreted for every packet at every programmable hop. Existing integration through P4sim provides realistic BMv2 execution inside ns-3, but ns-3 uses a sequential event core. Days is a recent multicore discrete-event network simulator whose lightweight packet representation intentionally carries no serialized protocol bytes. Integrating BMv2 with Days therefore requires more than a direct wrapper. BMv2 requires packet bytes and persistent per-switch state, whereas Days must retain control of time, routing, links, queueing, and parallel execution. We present P4-Days, an extension that embeds one BMv2 instance in each programmable Days switch. P4-Days lazily synthesizes protocol headers, executes P4 ingress and egress on opposite sides of the Days scheduler, and isolates mutable BMv2 state at the switch level. Across 16 cross-simulator scenarios, P4-Days matches ns-3+p4sim exactly in every scenario in which no output port is persistently saturated, a condition that is computable from the routing before a run. Where saturation makes the outcome depend on input arbitration, which neither simulator models, the two diverge; unmodified Days and P4-Days remain byte-identical in exactly those scenarios. P4-Days is 1.82-1.97x faster with stock BMv2 and 1.94-2.52x faster with logging-free BMv2. On a 1,280-switch fat-tree, P4-Days reaches a 5.19x speedup on 16 workers, whereas unmodified Days reaches 3.96x. These results show that programmable packet processing can be integrated without destroying Days' actor-level parallelism.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-09-24
DOI
https://doi.org/10.5281/zenodo.22936768
Primary Topic
Software-Defined Networks and 5G
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
preprint

P4-Days: Programmable Data-Plane Execution in a Multicore Discrete-Event Network Simulator

Taejoon Lee
Zenodo (CERN European Organization for Nuclear Research)
Software-Defined Networks and 5G
preprint

P4-Days: Programmable Data-Plane Execution in a Multicore Discrete-Event Network Simulator

Taejoon Lee
preprint en

Abstract

Packet-level simulation of P4-programmable networks is expensive because a software P4 pipeline must be interpreted for every packet at every programmable hop. Existing integration through P4sim provides realistic BMv2 execution inside ns-3, but ns-3 uses a sequential event core. Days is a recent multicore discrete-event network simulator whose lightweight packet representation intentionally carries no serialized protocol bytes. Integrating BMv2 with Days therefore requires more than a direct wrapper. BMv2 requires packet bytes and persistent per-switch state, whereas Days must retain control of time, routing, links, queueing, and parallel execution. We present P4-Days, an extension that embeds one BMv2 instance in each programmable Days switch. P4-Days lazily synthesizes protocol headers, executes P4 ingress and egress on opposite sides of the Days scheduler, and isolates mutable BMv2 state at the switch level. Across 16 cross-simulator scenarios, P4-Days matches ns-3+p4sim exactly in every scenario in which no output port is persistently saturated, a condition that is computable from the routing before a run. Where saturation makes the outcome depend on input arbitration, which neither simulator models, the two diverge; unmodified Days and P4-Days remain byte-identical in exactly those scenarios. P4-Days is 1.82-1.97x faster with stock BMv2 and 1.94-2.52x faster with logging-free BMv2. On a 1,280-switch fat-tree, P4-Days reaches a 5.19x speedup on 16 workers, whereas unmodified Days reaches 3.96x. These results show that programmable packet processing can be integrated without destroying Days' actor-level parallelism.

Zenodo (CERN European Organization for Nuclear Research)
University of Ottawa (CA)
Peace, Justice and strong institutions
Software-Defined Networks and 5G
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

P4-Days: Programmable Data-Plane Execution in a Multicore Discrete-Event Network Simulator — Taejoon Lee · Zenodo (CERN European Organization for Nuclear Research) (2026) | TGRS Research Map | TGRS