Browse Prior Art Database

Method and System for Enabling a PaaS System to Forecast Near-Term Service Consumption

IP.com Disclosure Number: IPCOM000243130D
Publication Date: 2015-Sep-16
Document File: 3 page(s) / 29K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for enabling a Platform as a Service (PaaS) system to forecast near-term service consumption based on current consumption of other services located within the PaaS.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 47% of the total text.

Page 01 of 3

Method and System for Enabling a PaaS System to Forecast Near -Term Service Consumption

Application architects are increasingly designing web applications through a composition of services or micro-services to increase fault tolerance and provide easy scalability. The micro-services provide functions to other components in a distributed application, and are augmented as needed. The distributed applications are frequently hosted in cloud environments to take advantage of cloud's large numbers of on-demand resources. However, current Platform as a Service (PaaS) providers do not provide an easy mechanism to predictively ramp up services to meet upcoming demand. Most auto-scaling services are reactive rather than proactive.

Disclosed is a method and system for enabling a Platform as a Service (PaaS) system to forecast near-term service consumption based on current consumption of other services located within the PaaS. The method and system enables a PaaS cloud to determine when application components or services of an application that are being hosted within the same PaaS cloud by analyzing service usage patterns over time. Once an association is determined, improvement manageability is possible through predictive scaling.

For example if the United States Supreme Court issues a key ruling on an important topic that results in a flash of social media posts which focus on sharing small amounts of information with multiple individuals. Shortly thereafter, a social media site that is general with more substantive content sees a large increase in traffic. Sometime after that, services composing a news site become busy as news publishers put out articles about the decision. Lastly, services that provide ability for users to comment on news articles becomes busy as people discuss the article on the news website. The pattern may repeat itself often, even if an initial stimulus to the first social service is different. If all of these services are hosted on a same PaaS infrastructure, the method and system discovers patterns and uses the patterns to predictively scale downstream services. The method and system is also applicable when two services both hosted on a PaaS cloud are separate entities and may not have an obvious connection.

In accordance with the method and system, a pattern discovery component analyzes historical service consumption previously persisted by a service consumption component. The pattern discovery component uses statistical methods to look for patterns that exist between service consumption of one service and near-term future service consumption of other services that may vary in both complexity and accuracy.

In one potential embodiment, service consumption metrics are stored in a map/reduce system. Periodically, jobs are run to reduce the service consumption information for various services into a summary for specific times. Then a second map/reduce job may look through that reduced data set to determine any potenti...