Feasibility study on distributed simulations of BGP

TL;DR

This paper investigates the feasibility of extending a large-scale distributed simulation platform, DRMSim, to support BGP routing protocol simulations on Internet-sized topologies, analyzing communication overhead and performance implications.

Contribution

It proposes a distributed extension of DRMSim for simulating BGP on large topologies and evaluates the communication overhead and performance feasibility.

Findings

Distributed simulation of BGP is feasible with manageable overhead.

Partitioning strategies impact communication costs significantly.

Expected additional simulation time remains acceptable for large topologies.

Abstract

The Autonomous System (AS)-level topology of the Internet that currently comprises 40k ASs, is growing at a rate of about 10% per year. In these conditions, Border Gateway Protocol (BGP), the inter-domain routing protocol of the Internet starts to show its limits, among others in terms of the number of routing table entries it can dynamically process and control. To overcome this challenging situation, the design but also the evaluation of alternative dynamic routing models and their comparison with BGP shall be performed by means of simulation. For this purpose, DRMSim, a Dynamic Routing Model Simulator, was developed that provides the means for large-scale simulations of various routing models including BGP. By means of this discrete-event simulator, execution of path-vector routing, e.g. BGP, and other compact routing models have been successfully performed on network topologies comprising more than ten thousand (abstract) nodes. However, to simulate dynamic routing schemes like BGP, DRMSim needs enhancements to support current Internet size (40k ASs) and even more by considering its evolution (up to 100k ASs). This paper proposes a feasibility study of the extension of DRMSim so as to support the Distributed Parallel Discrete Event paradigm. We first detail the possible distribution models and their associated communication overhead. Then, we analyze the communication overhead of such a distributed simulator by executing BGP on a partitioned topology according to different scenarios. Finally, we conclude on the feasibility of such a simulator by computing the expected additional time required by a distributed simulation of BGP compared to its sequential simulation.