Browse Prior Art Database

A Method and System for an Automated Partition Planner for Scalable Network Discovery and Root Cause Analysis

IP.com Disclosure Number: IPCOM000179820D
Original Publication Date: 2009-Feb-26
Included in the Prior Art Database: 2009-Feb-26
Document File: 3 page(s) / 69K

Publishing Venue

IBM

Abstract

Abstract: This article describes a network partition planner that enables distributed and concurrent discovery and root cause analysis over large scale networks.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 57% of the total text.

Page 1 of 3

A Method and System for an Automated Partition Planner for Scalable Network Discovery and Root Cause Analysis

A Method and System for an Automated Partition Planner For Scalable Network Discovery and Root Cause Analysis

      This article describes a network partition planner that enables distributed and concurrent discovery and root cause analysis over large scale networks.

Abstract:

P

 roblem Statement: As the size of a managed network increases, the cost of discovering it increases linearly and the cost of monitoring it increases super-linearly. Hence, network discovery and monitoring (e.g.: for root cause analysis RCA) is faced with serious performance and scalability bottlenecks. One potential solution to this problem is to partition the network into multiple management domains, each of which is separately discovered and managed. State of the art solutions require a network administrator to manually partition a large network into multiple management domains. However, manually partitioning a large network is prone to be cumbersome, error-some and sub-optimal. This invention describes a system and method to automatically partition a large network into multiple management domains using minimal input from the system administrator.

Technical Approach: The key idea is to leverage natural network partitions inherent in EGP and IGP networks to automatically identify (near) optimal management domains taking into account network technology and IP routing characteristics. In doing so, the partition planner attempts to rely on minimal manual input (e.g.: a set of managed AS numbers OR a set of EGP routers - at least one router per managed AS). Indeed the planner does not assume any a priori knowledge about the network topology. However, optimal partitioning depends on the network topology. The planner solves this bootstrapping problem by examining selected portions of the network using a technology driven approach. The goal here is to minimize the number of devices that need to be probed in order to partition the network into multiple management domains subject to the following three constraints: correctness,

                             erformance and placement. First, the partitions must satisfy the route closure property, that is, for any two devices in a partition, all routes between the two devices must be contained within the partition. This property is essential for the efficacy of monitoring and root cause analysis (RCA). Second, for performan...