Browse Prior Art Database

Component notification framework for sensing kernel events.

IP.com Disclosure Number: IPCOM000249095D
Publication Date: 2017-Feb-04
Document File: 3 page(s) / 32K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed in this publication is a research solution to address this issue of duplication by providing a generalized framework to deal with all the three aspects - sensing, effecting and knowledge management, which are just abstract entities in the MAPE-K control loop. Autonomic Computing (AC) has become extremely critical in managing large scale systems as it is becoming impractical to manage them physically. A general architectural model describes the MAPE-K (Monitor, Analyze, Plan and Execute guided by Knowledge) control loop with use of ‘Touchpoints’ and their managers implementing the sensor and effector behavior for the managed resources. With many Touchpoint managers in our system operating on common sensors and effectors it was essential to come up with a centralized framework that efficiently abstracted these entities for us and third party developers.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 33% of the total text.

1

Component notification framework for sensing kernel events .

Introduction

Disclosed in this publication is a research solution to address the issue of duplication by providing a generalized framework to deal with all the three aspects - sensing, effecting and knowledge management, which are just abstract entities in the MAPE-K control loop. Explosive growth in demand for business process automation, technology blend, diversity, governance, conformance, and most importantly market agility - are underpinning the need for high performance and adaptive solutions and applications. A business organization can no longer just accelerate current practices and rely on human intelligence to scale. To cope up with this limitation, their applications need to be autonomic or in other words self managing. One key aspect for an application to be autonomic is the ability to analyze the environment around it and invoke appropriate actions to optimize its performance and thereby maintaining user satisfaction.

Autonomic software suggests these properties as in self configuration, self-healing, self-optimization and self-protection be available in the applications. One of the key differentiating factors customers observe while buying an enterprise operating system is its performance, and the things which are relevant to it in our context is the ability of an operating system to support self-configuring and self-optimizing applications. To address this new frontier applications based on MAPE-K control loop was designed.

The MAPE-K control loop has integral components as - Manage, Analyze, Plan, Execute, Knowledge and further includes the concept of 'Touchpoints' which implement sensing and effecting of the autonomic system. When an autonomic component in an operating system was being developed it was realized that the effort was duplicated in most places every time autonomicity was introduced into the operating system. Key things that were being duplicated were the sensing, effecting and the knowledge part of the MAPE-K control loop. For example, sensing was being duplicated by two applications that needed to be informed by a change in the very same kernel parameter and then the message needed to be delivered to the applications via some kernel-to-users space communication mechanism (callbacks). If it wasn’t complete duplication, sometimes it was required to re-use the same kernel-to-user space mechanisms or sometimes needed to just call something in the kernel itself.

To address this issue an idea of a generalized framework in kernel that can be used by the operating system and third party developers was designed. It also aims to address the needs of applications which needed to be autonomic covering the aspects of sensing, effecting and knowledge management. Even though our main goal is to architect a unified framework which can manage our sensors, effectors and the associated knowledge enabling a scalable abstraction to the programmers, in this publication a...