Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

DISTRIBUTED TRAFFIC TRACING ACROSS DEVICES IN LARGE NETWORKS

IP.com Disclosure Number: IPCOM000248007D
Publication Date: 2016-Oct-17
Document File: 6 page(s) / 2M

Publishing Venue

The IP.com Prior Art Database

Related People

Adam Gargulak: AUTHOR [+4]

Abstract

A distributed packet tracer (DPT) provides a unified, programmable, and platform-independent infrastructure. The infrastructure enables efficiently finding, tracing, and analyzing specific packet flows in large distributed computing networks. The infrastructure may be integrated with a wide range of hardware/software platforms.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 30% of the total text.

Page 01 of 6

DISTRIBUTED TRAFFIC TRACING ACROSS DEVICES IN LARGE NETWORKS

AUTHORS:

   
Adam Gargulak
Naveen Kumar Krishnappa Srinivas Pitta
Thomas Denotte

CISCO SYSTEMS, INC.

ABSTRACT

    A distributed packet tracer (DPT) provides a unified, programmable, and platform-independent infrastructure. The infrastructure enables efficiently finding, tracing, and analyzing specific packet flows in large distributed computing networks. The infrastructure may be integrated with a wide range of hardware/software platforms.

DETAILED DESCRIPTION

    Debugging of application traffic flow issues in large data center (DC) networks are difficult problems to resolve. Currently, debugging is carried out manually by network operators using multiple platform-dependent tools (e.g., flow tracing and analysis such as packet drop, flood, and latency) to determine the flow path and identify the device that poses the issue in the network and user interfaces, such as a command-line interface (CLI). Such debugging need to be executed on every device until exact path is found, that include hop by hop path analysis. The manual approach is extremely slow due to the increasing size and complexity of DC networks as well as the constantly changing underlying platforms. As a result, such troubleshooting can span a few hours to days.

    DC networks may consist of relatively large networks with different types of networking devices and multiple redundant paths. Due to such complexity, extensive time and operational resources are required to find a specific traffic path/flow/device to triage problem flow. These delays cause prolonged time to resolution (TTR). Platform- dependent setups cause individual users to interface for each platform/application specific

Copyright 2016 Cisco Systems, Inc.

1


Page 02 of 6

integrated circuit (ASIC). This leads to an inability to effectively integrate with the orchestrator/controller. Current tools lack a unified programmable interface and the ability to integrate with massive scale network orchestrators/controllers.

    Figure 1 below shows a current triaging (static) model. The model is platform dependent (e.g., CLI/application program interface (API) is different for each ASIC vendor). As a result, there are interoperability issues between the platform/ASIC and the orchestrator. In this model, there is no unified programmable interface, an inability to integrate with massive scale network orchestrator/controllers, and an inability to isolate the problem to one device in a dynamic way.

Figure 1

    There is a need for an automated and platform-independent tool with a unified interface. In particular, an innovative approach is needed to effectively track traffic flow path and its state across complex networks without impacting traffic convergence time and system performance.

    As provided herein, a specific traffic flow path can be detected without disturbing the network or overloading the system. Unlike certain tools (e.g., traceroute) which inject a packet or platform-d...