Table of Contents
Fetching ...

LCDN: Providing Network Determinism with Low-Cost Switches

Philip Diederich, Yash Deshpande, Laura Becker, Wolfgang Kellerer

TL;DR

LCDN addresses the challenge of providing end-to-end deterministic networking for real-time applications on low-cost Ethernet hardware. It introduces a centralized control plane that uses Deterministic Network Calculus to model delays, VLAN-based source routing over multiple spanning trees, and per-class SPQ/IWRR scheduling, coupled with end-host Token-Bucket Filtering to enforce arrival curves. Key results show deterministic switch behavior on inexpensive hardware (processing time capped at about 4.15 µs, 8 queues, 3.5 µs queueing overhead) and robust end-host performance (TBF deviations often below 1.88%), validated on a Raspberry Pi. While noting limitations like VLAN scalability and lack of TAS on budget switches, the work demonstrates practical, cost-effective deterministic networking suitable for real-time testbeds and CPS deployments.

Abstract

The demands on networks are increasing at a fast pace. In particular, real-time applications have very strict network requirements. However, building a network that hosts real-time applications is a cost-intensive endeavor, especially for experimental systems such as testbeds. Systems that provide guaranteed real-time networking capabilities usually work with expensive software-defined switches. In contrast, real-time networking systems based on low-cost hardware face the limitation of lower link speeds. This paper fills this gap and presents Low-Cost Deterministic Networking (LCDN), a system designed to work with inexpensive, common off-the-shelf switches and devices. LCDN works at Gigabit speed and enables powerful testbeds to host real-time applications with strict delay guarantees. This paper also provides an evaluation of the determinism of the switch and a Raspberry Pi used as an end device to demonstrate the applicability of LCDN on inexpensive low-power reduced capacity apparatus.

LCDN: Providing Network Determinism with Low-Cost Switches

TL;DR

LCDN addresses the challenge of providing end-to-end deterministic networking for real-time applications on low-cost Ethernet hardware. It introduces a centralized control plane that uses Deterministic Network Calculus to model delays, VLAN-based source routing over multiple spanning trees, and per-class SPQ/IWRR scheduling, coupled with end-host Token-Bucket Filtering to enforce arrival curves. Key results show deterministic switch behavior on inexpensive hardware (processing time capped at about 4.15 µs, 8 queues, 3.5 µs queueing overhead) and robust end-host performance (TBF deviations often below 1.88%), validated on a Raspberry Pi. While noting limitations like VLAN scalability and lack of TAS on budget switches, the work demonstrates practical, cost-effective deterministic networking suitable for real-time testbeds and CPS deployments.

Abstract

The demands on networks are increasing at a fast pace. In particular, real-time applications have very strict network requirements. However, building a network that hosts real-time applications is a cost-intensive endeavor, especially for experimental systems such as testbeds. Systems that provide guaranteed real-time networking capabilities usually work with expensive software-defined switches. In contrast, real-time networking systems based on low-cost hardware face the limitation of lower link speeds. This paper fills this gap and presents Low-Cost Deterministic Networking (LCDN), a system designed to work with inexpensive, common off-the-shelf switches and devices. LCDN works at Gigabit speed and enables powerful testbeds to host real-time applications with strict delay guarantees. This paper also provides an evaluation of the determinism of the switch and a Raspberry Pi used as an end device to demonstrate the applicability of LCDN on inexpensive low-power reduced capacity apparatus.
Paper Structure (7 sections, 4 figures, 1 table)

This paper contains 7 sections, 4 figures, 1 table.

Figures (4)

  • Figure 1: Overview of the LCDN architecture. The switches and end hosts on the user plane connect to the centralized LCDN manager. The manager has four different components for different tasks: the DNC agent checks if a new requested flow can be embedded given the current state of the network by iterating over all paths with the routing agent. It also provides other flows which need to be reconfigured. The topology manager keeps track of the network graph and any connections/disconnections while the flow dispatcher handles communication with the end hosts and reconfigures paths if needed.
  • Figure 2: Example assignment of VLAN Identifiers to multiple spanning trees. End-hosts are excluded as they have one connection to a switch and, hence, are already loop-free. 2 VLANs share every single hop path while every 2-hop path has a unique VLAN ID.
  • Figure 3: The measured processing times of the switch in various scenarios.
  • Figure 4: End host measurements comparing the configured rate of TBF and the measured rate at the receiver. The rate test varies the TBF rate and uses a constant burst size of 1542B (layer 1). The burst size test uses a constant TBF rate of 3Mbps and varies the burst size, which is equal to the application packet size on L1.