Browse Prior Art Database

Method and System for Diagnosing Anomalies and Providing Cloud Performance Identity

IP.com Disclosure Number: IPCOM000249318D
Publication Date: 2017-Feb-16
Document File: 4 page(s) / 404K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for distinguishing faults with a globally direct impact from faults with a locally direct impact, which is critical for efficiently diagnosing performance anomalies in shared hosting infrastructures.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

1

Method and System for Diagnosing Anomalies and Providing Cloud Performance Identity

The current incident management solutions typically react to issues or complaints raised by a user with respect to performance deterioration or failure. The symptoms associated with the performance deterioration or failure are utilized to drill down into the systems, identify the issue and to fix the issues. For issues that are well known, standard workflows are developed to accelerate the remediation. However, with the introduction of cloud technology, the pace of upgrades both at an environment level and at a services and applications level makes it difficult to efficiently maintain a meaningful collection of remediation workflows.

The currently available solutions use Auto-Regressive models to detect time-invariant relationships for monitoring data and analyze broken invariants during runtime in order to localize the fault. A principle component analysis and an independent component analysis is used along with outlier detection to automatically detect faulty components in large-scale systems.

Further, a white-box approach for analyzing performance bugs works by manually checking the patches of known problems and then build an efficiency rule-based checker to identify previously unknown performance problems in deployed software. This approach automatically identifies and ranks anomalous function calls using robust principle component analysis along with white-box instrumentation to provide fine-grained performance debugging information to developers. However, it is difficult to obtain the source code access for production systems running inside the virtualized hosting infrastructures.

Thus, there is a need for a method and system that addresses the above mentioned disadvantages and for efficiently diagnosing performance anomalies in shared hosting infrastructures.

Disclosed is a method and system for distinguishing faults with a globally direct impact from faults with a locally direct impact, which is critical for efficiently diagnosing performance anomalies in shared hosting infrastructures.

From a provider end, the method and system builds a catalog of platform solution patterns mapped to service management suppor...