Browse Prior Art Database

Infrastructure Autoprovision and Learning Based on User's Feedback

IP.com Disclosure Number: IPCOM000240390D
Publication Date: 2015-Jan-28
Document File: 2 page(s) / 76K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system to improve infrastructure resource provisioning based on user feedback. The novel contribution is a mechanism to survey users and obtain and analyze feedback, and then automatically adjust resource allocation in order to improve performance and/or reduce costs.

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

Page 01 of 2

Infrastructure Autoprovision and Learning Based on User's Feedback

Infrastructure resource provisioning has to account for all Service Level Agreements (SLAs) and Service Level Objectives (SLOs). Existing systems contain a number of thresholds and monitors to gauge performance and ensure that a certain level is achieved.

However, in real systems, the metric of interest is often user-perceived responsiveness. While SLAs and SLOs are crafted in an attempt to satisfy users, such guidelines are often ineffective. The SLAs and SLOs can be overly aggressive, providing a level of response that is costly and unnecessary, or can be under-aggressive, leading to customer dissatisfaction.

In cases where users matter, user feedback provides the most accurate way to validate performance. This form of validation then allows the infrastructure to learn and adapt the changing needs at the application level as well as ensure that an expected level of performance is achieved. Using this method requires users to participate in a given survey for which random sampling is employed to ensure that the results are not biased.

A system is needed to improve infrastructure resource provisioning based on user

feedback.

The novel contribution is a mechanism to survey users and obtain feedback for analysis. If user feedback trends negatively, then the cloud infrastructure determines the bottleneck (perhaps using monitoring or more specific user questions) and automatically provisions more res...