Browse Prior Art Database

Method and System to adjust application deployment dynamically in heterogeneous environment for workload optimization

IP.com Disclosure Number: IPCOM000223358D
Publication Date: 2012-Nov-20

Publishing Venue

The IP.com Prior Art Database

Abstract

This article provides a method and system to optimize workload by adjusting application deployment dynamically. This idea is different with existing load balance technology. In order to regroup application deployment dynamically in heterogeneous environment, Corresponding repository design and analysis engine are the key technologies in this idea. The analysis engine can generate the command to control application deployment according to the monitoring information and policy definition like SLA. As a result, this idea can improve the resource utilization and help to ensure all applications satisfy its QOS requirement

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 43% of the total text.

Page 01 of 10

Method and System to adjust application deployment dynamically in heterogeneous environment for workload optimization

Cloud platform offers elastic computing capacities. IT resources can be allocated automatically based on monitoring result or manually by invoking corresponding service. But this kind of auto scaling is usually in infrastructural level and doesn't care about what applications are running on it and what their QOS requirements are. And in an enterprise, they may have multiple applications deployed in different runtime environments which make it more difficult to manage and improve the resource utilization .

Priori arts

After investigating several existing cloud computing and load balance technologies, we find that they have common shortcomings as below:

Shortcomings:


1) Applications deployed in runtime instance keep unchanged, may cause low resource utilization.


2) The auto scaling with whole runtime instance may take a lot of time and can't response in time


3) Lack of advanced pattern analysis and self reinforcement optimization

The main idea includes:


1. Select suitable target runtime instance to add more application deployment based on SLA definition and metadata information


2. Adjust application deployment dynamically based on complementarily relationship

Here are the sample scenarios of this idea:


1) Add more application deployment

Once the work load for the runtime is very high, and need to scale in, then we will analysis the work load, and pick up one or more applications to deploy them in another runtime, so that requests can be routed to these new deployments


2) Adjust application deployment dynamically based on complementarily relationship

Different applications have different runtime information , and not all the applications have high throughput for all time , they may have complementarily relationship as shown in the picture below. Application A and Application B have different throughput from the time , when application A's throughput is high, application B's throughput is low, so that we can group these applications with complementarily relationship together and deploy together, as a result, the IT resource will always have high utilization.

1


Page 02 of 10

Advantage of this invention:


1) Achieve better resource usage by adjust application deployment automatically


2) Auto scaling in application level which will save time than in infrastructure level


3) Application level information such as SLA/SLD policy is used for auto scaling

Description:

Since companies may have multiple kinds of runtime instances running on different platforms, and runtime instances have dynamic workload, so it is difficult to select suitable target runtime instance to add more application deployment automatically. The detailed solution looks as below:
System architecture

2


Page 03 of 10


1. Domain:


a. Each domain contains same type of runtime environment

b. It can only contain one runtime instance or multiple runtime instances wh...