[Paper Review] Topology Discovery Using Cisco Discovery Protocol

Sergio R. Rodriguez|ArXiv.org|Jul 13, 2009

Software System Performance and Reliability3 references3 citations

TL;DR

This paper presents a method to discover the physical topology of Cisco-based networks using Cisco Discovery Protocol (CDP), demonstrating that CDP can reveal links in Forwarding Block state—information not accessible via STP or MIB FDB. The approach leverages CDP's device-to-device discovery to reconstruct a complete, accurate network map, offering a practical solution for topology discovery in proprietary Cisco environments.

ABSTRACT

In this paper we address the problem of discovering network topology in proprietary networks. Namely, we investigate topology discovery in Cisco-based networks. Cisco devices run Cisco Discovery Protocol (CDP) which holds information about these devices. We first compare properties of topologies that can be obtained from networks deploying CDP versus Spanning Tree Protocol (STP) and Management Information Base (MIB) Forwarding Database (FDB). Then we describe a method of discovering topology of CDP-based networks. Our experiments show that the physical topology of the network including links that are in Forwarding Block state can be discovered.

Motivation & Objective

• To address the challenge of discovering accurate physical network topology in proprietary Cisco-based networks.
• To compare the topological information obtainable from CDP, Spanning Tree Protocol (STP), and MIB Forwarding Database (FDB).
• To develop and validate a method for reconstructing the full physical topology using CDP data.
• To demonstrate that CDP can detect links in Forwarding Block state, which are invisible to STP and FDB.
• To provide a practical, scalable solution for network administrators to map network structures without relying on standard SNMP or STP mechanisms.

Proposed method

• The method collects CDP advertisements from network devices, extracting neighbor information such as device ID, port, and capabilities.
• It constructs a directed graph from CDP data, modeling bidirectional links based on mutual CDP advertisements.
• The approach correlates CDP entries with interface states to identify links in Forwarding Block state.
• It uses a merging algorithm to resolve inconsistencies and build a consistent topology map from multiple CDP reports.
• The method validates topology accuracy by comparing discovered links with known network configurations and device logs.
• It leverages existing CDP infrastructure without requiring additional protocols or configuration changes.

Experimental results

Research questions

• RQ1 Can CDP provide a more complete view of network topology than STP or MIB FDB?
• RQ2 What portion of the physical topology, including blocked links, can be discovered using CDP?
• RQ3 How accurate and consistent is the topology reconstructed from CDP data across diverse Cisco network configurations?
• RQ4 Can CDP-based topology discovery be reliably automated and scaled in real-world networks?
• RQ5 What are the limitations of CDP compared to other discovery mechanisms like STP or SNMP?

Key findings

• CDP can discover the complete physical topology of a network, including links in Forwarding Block state, which are not visible through STP or MIB FDB.
• The topology reconstruction method using CDP achieves high accuracy, with experimental results confirming correct mapping of all physical links in test networks.
• CDP-based discovery outperforms STP and MIB FDB in capturing all active and blocked links, providing a more comprehensive view of network structure.
• The method successfully identifies bidirectional links by validating mutual CDP advertisements between neighboring devices.
• The approach is scalable and practical, as it relies solely on existing CDP functionality without requiring protocol extensions or additional configuration.
• The study confirms that CDP is a viable and underutilized source for complete network topology discovery in Cisco environments.