Browse Prior Art Database

A system and method to manage application scaling in cloud environment

IP.com Disclosure Number: IPCOM000244220D
Publication Date: 2015-Nov-24
Document File: 4 page(s) / 83K

Publishing Venue

The IP.com Prior Art Database

Abstract

Scaling is an important capability provided in a cloud environment for applications to expand or shrink themselves according to some pre-defined rules. Currently, the scaling in a cloud environment is implemented in an out-of-box way and is coarse-grained. This disclosure provides a system and method allowing applications to re-evaluate whether to scale besides the routine checking is passed.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 4

A system and method to manage application scaling in cloud environment

Scaling is an important capability provided in a cloud environment for applications to expand or shrink themselves according to some pre-defined rules, by which the allocation of cloud resources can be optimized and thus IT expenses are saved. Currently, the scaling in a cloud environment is implemented in an out-of-box way and is coarse-grained.

This disclosure provides a system and method allowing applications to re-evaluate whether to scale besides the routine checking is passed. This new method introduces state analyzer and state evaluator, which will onlybe activated when the periodically collected metrics exceeds the threshold. The new functions dissociate the heavy part from the scaling determination process, and also improve the accuracy of scaling operations.

FIG 1 illustrates the architecture overview of triggering scaling operations.

FIG 1. Architecture Overview

An application may have one or more components. Theterm "Component" in the figure means the minimum deployment unit that cloud providers can allocate resources to. In IaaS, it could be a VM, and in PaaS, it cloud be a container. The scaling operations are performed at the component level.

1



Page 02 of 4

Metrics are collected for each component instance, and persisted in Metrics Repository. Based on thresholds/rules defined by applications, Metrics Rule Analyzer assesses metrics and may get the result that a particular instance is in a "suspectable" status. Then the Metrics Rule Analyzer requests Cluster State Evaluator to perform an ad-hock check to get the overall state of the component cluster.

A Cloud provider provides some predefined componentstates and corresponding operations, and publishesthem to applications. For a multi-component application, the state is determined at the component level, the same level as scaling operations are performed at. Applications can add state check logic in the state analyzer module, which returns the current state of the component instance on Cluster State Evaluator's request. FIG 2 illustrates an example of the definition of predefined state-operation mapping.

STATE

Operation

GOOD NA FAIL Provision a new component instance and release the FAIL one

BUSY

Provision a new component instance and add it...