Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Predicting the resource requirements of applications to enable optimal deployment of the applications within a cloud

IP.com Disclosure Number: IPCOM000238006D
Publication Date: 2014-Jul-25
Document File: 4 page(s) / 92K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed here is an approach for predicting the "applications resource usage life pattern" for an application that is busy following business events that are known in advance to the business, but do not occur at regular intervals.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 49% of the total text.

Page 01 of 4

Predicting the resource requirements of applications to enable optimal deployment of the applications within a cloud

In a cloud computing environment, where the cloud is composed of 2 or more hardware devices to which applications are deployed, at the time of deployment of an application, the cloud must decide which hardware device or set of devices the application should be deployed to.

This deployment decision should allow optimal use of the hardware devices by multiple applications whilst ensuring that an individual hardware device is not overloaded, degrading the performance of applications it is hosting.

This decision could be made manually by a user, but the drawback is that the user has to be involved and may make mistakes. A round robin algorithm could be used, but that could overload some hardware devices whilst leaving others under utilized.

A more advanced option might be to observe the utilization of the hardware resources at the time the application is to be deployed, and to select a hardware device that is least utilized. The drawback of this approach is that this is a point in time observation. If a hardware device was running applications that were busy between 9am and 5pm, then at 8.45 am it might seem sensible to deploy an additional application to that hardware device, but at 9am that decision may prove to have been a bad one.

This problem can be partially solved by recording data about the resource usage of each application in the cloud, looking for patterns in that data and using the "applications resource usage pattern" to determine which hardware device is best suited to run the application. For example, application A may be heavily utilized between 7am and 9am, and between 7pm and 9pm. Application B may be heavily utilized between 9am and 5pm, so applications A and B could be deployed to the same hardware once their "application resource usage patterns" are understood.

The drawback of this approach is that not all applications will have workload patterns that can be predicted using this historical resource usage data. For example, applications that are busy following business events that are known in advance to the business, but do not occur at regular intervals.

By combining the knowledge of upcoming business events, together with historical business data about similar past business events and other information streams (such as judging the popularity of an event by monitoring social media), a prediction can be made as to the resources necessary for the application to handle the business event at the time the event occurs and during the lifetime of the event. It is the concept of predicting the applications resource usage pattern over the lifetime of the event (the applications "life pattern") and using that information to determine where to best place applications on hardware resources that we discuss further here .

The advantage of using this predictive approach is that optimal use of the clouds resource...