Browse Prior Art Database

System for consumer VM self-monitoring and calibration of performance results

IP.com Disclosure Number: IPCOM000235660D
Publication Date: 2014-Mar-18
Document File: 2 page(s) / 42K

Publishing Venue

The IP.com Prior Art Database

Abstract

Software running on virtual machines can be monitored and profiled for performance even if the user of the virtual machine (VM) has no visibility into the hypervisor that hosts the VM. A benchmark test can be run before, during, and after a test of software in order to calibrate the performance of the VM, and compare it to past or future tests.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 37% of the total text.

Page 01 of 2

System for consumer VM self -monitoring and calibration of performance results

The virtualization technology has been adopted in a variety of software testing areas by the virtue of its rapid provisioning of consistent testing systems, accelerated product deployment time, reduced testing costs and improved productivity. Despite its many advantages, it is not widely used in performance testing because an isolated test bed is hard to be obtained in a shared virtualized environment. The performance running under one virtual machine can be affected by other virtual machines running on the same virtualization platform and the testing results are not accurate or reliable. The invention proposed here enables performance tester to achieve a reliable testing results while fully leveraging the virtualization technology advantages in their testing.

The essence of this invention is one of a client that is running multiple VMs on a larger cloud service (Amazon*). This invention assumes that there is a set of proprietary data that will not be shared with client, and will not be available for comparison. (router info, network info, ESXI Server CPU utilization info, ESXI Server overall memory usage info)

This invention attempts to gather data from client vantage point that will allow the client to generate useful, predictive modeling.

End users like performance testers usually do not have access to administrative and statistics data about a virtualization server (e.g. VMWare ESX**) such as resource availability and usage. This invention is to introduce a benchmark testing running on a single virtual machine and enables performance testers to determine the best time to run their testing under a certain virtualized testing environment, validate and calibrate (if possible) their testing results, thus obtaining a reliable and accurate performance results.

Configuration options:
Run "benchmark" (sniff) test before/after product test

There will be a set of routines that are part of the benchmark test that are lightweight (vmstat, iostat,). There will be a set of more expensive scenarios that can be run before and after benchmark test).

Execute "benchmark" (sniff) test before/during/after test

Prior Art Item

Description

Difference with this idea

Performance testing in a cloud environment -

http://www.google.com/patents/US2012 0311128?dq=performance+testing+resu lt+in+a+cloud&hl=en&sa=X&ei=mJUbUr TSK9C3qAG70IHwAQ&sqi=2&pjf=1&ve d=0CDQQ6AEwAA

US 2012/0311128 is concerned with how to generate enough load in a cloud environment to cause an elastic application to be provisioned across multiple servers. The application may initially run on one server (or small set of servers), but if it gets enough load, it will be provisioned across more servers. US 2012/0311128 describes a method to cause this load to be generated.

This idea is not concerned with ensuring that a cloud-enabled application is tested in multiple elastic configurations. Rather, this invention i...