Browse Prior Art Database

Method and System for Deploying Graph Applications in Distributed Cloud Networks

IP.com Disclosure Number: IPCOM000247344D
Publication Date: 2016-Aug-25
Document File: 3 page(s) / 130K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for deploying graph applications in distributed cloud networks.

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

Page 01 of 3

Method and System for Deploying Graph Applications in Distributed Cloud Networks

Application deployment mechanisms in cloud computing systems deploy application

workloads to cloud in order to optimize performance. Application deployment is a crucial component of a cloud computing system because efficient allocation of resources yields low cost operation and enhanced user experience. Furthermore, it contain direct implications on the cost of operating the data centers and the charging costs to the users. The problem that arises when deploying graph application

workloads on distributed cloud networks is application deployments that may result in network routing loops. Such deployments result in high application data delays, reduced overall graph application performance (e.g. real time applications where computations on a vertex depend on data from previous vertices). This problem becomes more acute in distributed cloud environments (Internet/WAN clouds, CDNs, core/edge mobile clouds, etc.) where network delay is to be accounted for. There needs a method and system that can deploy graph applications in either distributed or cloud computing networks.

Disclosed is a method and system for deploying graph applications in distributed cloud networks. The method and system deploys graph applications by incorporating the application data direction and network layer routing in the application deployment decision. Graph applications are deployed in distributed cloud network by using plurality of modules such as, a storage module comprising graph representations of physical topology and its routing topologies/applications, a monitoring module comprising a physical resource monitor and an application monitor, a placement module for computing application placement, and a control module comprising an application deploy and application queue.

In accordance with the method and system, as illustrated in figure 1, an application graph G is sent to a server that comprises knowledge of the cloud network state. The application graph include guidelines such as, but not limited to, which host locations in the data center some operations of the application need to be performed and where the incoming application data traffic comes from. The application graph G enables identification of the dominant directions of data flow and computations in the application graph. Subsequently, the dominant directions are provided by one or more of programming models or applications. Thereafter, in applications where the graph is not directed, dominant directions can be extracted by profiling the application traffic on the edges of the graph for obtaining the output of the step which is a directed application graph G'.

1


Page 02 of 3

Figure 1

In...