Browse Prior Art Database

Apparatus to arrive at an optimized topology of cloud operating system services.

IP.com Disclosure Number: IPCOM000241520D
Publication Date: 2015-May-08
Document File: 5 page(s) / 108K

Publishing Venue

The IP.com Prior Art Database

Abstract

The distributed cloud software architectures come with a scale out functionality for it's services. These services are NOT the services to be deployed on the cloud but are the processes that constitute the IAAS/PAAS service. For example openstack constitutes of multiple services like Nova-api, Cinder, Glance etc. A scale out functionally typically involves distributing the services constituting the cloud operating system into multiple nodes. In this context - the openstack services can be put out on different nodes. This is done to ensure scalability of the cloud operating system. When services are scaled out –a broker service is used to tie all the distributed services. However simply scaling out the services for better computation has it's own drawbacks. There is a considerable performance impact during inter service communications that are done across multiple nodes. Maintenance cost increases as the systems scales out and the performance overhead increases. This article tries to solve the problem of scaling out of cloud operating services by devising an algorithm for intelligent service management. The key feature of this disclosed mechanism is to provide a way by which a topology defining the services constituting cloud operating system can be derived. The topology of a cloud operating system services takes into account affinity of various services and models them to be placed in a more systematic/efficient way for better throughput. The affinity model is calculated using a static and a dynamic analysis of API definition/usages of each service. The affinity model also takes into consideration a dependency graph comprising of each of the cloud operating system services and in turn outputs the collocation suggestions. The dependency graph shows the nature of dependency of each of the services with respect to others. A payload size is taken into consideration –that defines the size of the data transaction across multiple services. With the above in consideration. Following are the unique features: 1. An apparatus to draw a topology of cloud operating system services to help the system admin on deciding the distribution of services - suited to optimize service usability. 2. A score based mechanism by which a cloud operating system service activity can be measured that denotes resource usability. 3 . A dependency graph that can be used to determine service affinities which in turn determines the topology of services.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 26% of the total text.

Page 01 of 5

Apparatus to arrive at an optimized topology of cloud operating system services .

The distributed cloud software architectures come with a scale out functionality for it's services. These services are NOT the services to be deployed on the cloud but are the processes that constitute the IAAS/PAAS service. For example openstack constitutes of multiple services like Nova-api, Cinder, Glance etc. A scale out functionally typically involves distributing the services constituting the cloud operating system into multiple nodes. In this context - the openstack services can be put out on different nodes. This is done to ensure scalability of the cloud operating system. When services are scaled out -a broker service is used to tie all the distributed services. However simply scaling out the services for better computation has it's own drawbacks. There is a considerable performance impact during inter service communications that are done across multiple nodes. Maintenance cost increases as the systems scales out and the performance overhead increases.

This article tries to solve the problem of scaling out of cloud operating services by devising an algorithm for intelligent service management.

The key feature of this disclosed mechanism is to provide a way by which a topology defining the services constituting cloud operating system can be derived. The topology of a cloud operating system services takes into account affinity of various services and models them to be placed in a more systematic/efficient way for better throughput. The affinity model is calculated using a static and a dynamic analysis of API definition/usages of each service. The affinity model also takes into consideration a dependency graph comprising of each of the cloud operating system services and in turn outputs the collocation suggestions. The dependency graph shows the nature of dependency of each of the services with respect to others. A payload size is taken into consideration -that defines the size of the data transaction across multiple services. With the above in consideration.

Following are the unique features:

1. An apparatus to draw a topology of cloud operating system services to help the system admin on deciding the distribution of services - suited to optimize service usability.

2. A score based mechanism by which a cloud operating system service activity can be measured that denotes resource usability.

3 . A dependency graph that can be used to determine service affinities which in turn determines the topology of services.

A distributed cloud operating system constitutes of multiple services. For example - Openstack  has Glance which is used for image management, neutron used for network, nova­api service to  interface with the core nova services etc. These services require interoperability, which is  achieved by making RPC or REST API calls.  Most of these models are built on the philosophy  of ...