Demand forecasting system for data center, demand forecasting method and recording medium with a demand forecasting program recorded thereon

• US 20060218278A1
• Filed: 07/29/2005
• Published: 09/28/2006
• Est. Priority Date: 03/24/2005
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Yasaki, Masatomo, Uyama, Masashi, Kohda, Youji
• Granted Patent: US 8,260,921 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: H04L 41/147 for predicting network behaviour
• CPC Class Codes: H04L 41/5064 Customer relationship management
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]
• CPC Class Codes: H04L 67/535 Tracking the activity of the user network monitoring arrangements H04L43/00; recording of computer activity G06F11/34
• CPC Class Codes: H04L 69/02 Protocol performance

Abstract

A demand forecasting system and a demand forecasting method capable of forecasting demands for IT resources in a data center accurately are provided. The demand forecasting system includes: a classification unit that classifies a plurality of users who use IT resources in a data center into a plurality of user'"'"'s behavior classes; an event characteristics model generation unit that generates an event characteristics model representing a variation over time in a usage state of the IT resources caused by an event and the number of the users moving among the user'"'"'s behavior classes caused by the event; an event influence calculation unit that calculates a predicted value of the variation over time in a usage state of the IT resources and a predicted value of the number of the users moving among the user'"'"'s behavior classes that will be caused by an event scheduled to be held in the future using the event characteristics model; and a demand forecasting unit that calculates a variation in load on the IT resources.

Claims

1. A demand forecasting system, comprising:

a classification unit that classifies a plurality of users who use IT resources in a data center into a plurality of user'"'"'s behavior classes based on behavior patterns of the users to use the IT resources;

an event characteristics model generation unit that generates an event characteristics model based on a record of usage by the users in the past, the event characteristics model representing a variation over time in a usage state of the IT resources caused by a past event for each user'"'"'s behavior class and the number of the users moving among the user'"'"'s behavior classes caused by the event;

an event influence calculation unit that calculates a predicted value of the variation over time in a usage state of the IT resources for each user'"'"'s behavior class and a predicted value of the number of the users moving among the user'"'"'s behavior classes that will be caused by an event scheduled to be held in the future using the event characteristics model; and

a demand forecasting unit that calculates a variation in load on the IT resources based on the predicted values of the variation over time and the number of the users moving that are calculated by the event influence calculation unit.

2. The demand forecasting system according to claim 1, wherein the classification unit classifies the users into the plurality of user'"'"'s behavior classes based on the behavior patterns of the users represented with a time between occurrence of the event and starting of usage of the IT resources by the users and frequencies of the users using the IT resources.

3. The demand forecasting system according to claim 1, further comprising a regular variation calculation unit that calculates a regular variation in a usage state of the IT resources for each user'"'"'s behavior class except for event attributable variations,

wherein the demand forecasting unit calculates the variation in load based on the variation calculated by the regular variation calculation unit and the predicted values of the variation over time and the number of the users moving calculated by the event influence calculation unit.

4. The demand forecasting system according to claim 1,

wherein the event characteristics model generation unit generates a plurality of event characteristics models,

the event influence calculation unit calculates the predicted values of the variation over time and the number of the users moving using one of the plurality of event characteristics models, and

the demand forecasting system further comprises a correction unit that collates an initial usage record after the occurrence of the scheduled event with the plurality of event characteristics models and corrects the predicted values of the variation over time and the number of the users moving calculated by the event influence calculation unit using an event characteristics model that is judged by the collating to have the closest resemblance to the usage record.

5. The demand forecasting system according to claim 1, wherein the demand forecasting unit calculates a variation in load on the IT resources for each user'"'"'s behavior class in accordance with each behavior pattern of the user'"'"'s behavior class, and synthesizes the variations in load of the respective user'"'"'s behavior classes.

6. A demand forecasting method that forecasts demands for IT resources in a data center using a computer that is accessible to usage recording data on which usage of the IT resources by a plurality of users is recorded, the method comprising the steps of:

a classification step in which a classification unit included in the computer classifies, based on the usage recording data, the plurality of users who use the IT resources in the data center into a plurality of user'"'"'s behavior classes according to behavior patterns of the users to use the IT resources;

an event characteristics model generation step in which an event characteristics model generation unit included in the computer generates, based on the usage recording data, an event characteristics model representing a variation over time in a usage state of the IT resources caused by a past event for each user'"'"'s behavior class and the number of the users moving among the user'"'"'s behavior classes caused by the event;

an event influence calculation step in which an event influence calculation unit included in the computer calculates a predicted value of the variation over time in a usage state of the IT resources for each user'"'"'s behavior class and a predicted value of the number of the users moving among the user'"'"'s behavior classes that will be caused by an event scheduled to be held in the future using the event characteristics model; and

a demand forecasting step in which a demand forecasting unit included in the computer calculates a variation in load on the IT resources based on the predicted values of the variation over time and the number of the users moving that are calculated at the event influence calculation step.

7. A computer readable recording medium having stored thereon a demand forecasting program executable to perform the processes of:

a classification process that classifies a plurality of users who use IT resources in a data center into a plurality of user'"'"'s behavior classes based on behavior patterns of the users to use the IT resources;

an event characteristics model generation process that generates an event characteristics model based on a record of usage by the users in the past, the event characteristics model representing a variation over time in a usage state of the IT resources caused by a past event for each user'"'"'s behavior class and the number of the users moving among the user'"'"'s behavior classes caused by the event;

an event influence calculation process that calculates a predicted value of the variation over time in usage of the IT resources for each user'"'"'s behavior class and a predicted value of the number of the users moving among the user'"'"'s behavior classes that will be caused by an event scheduled to be held in the future using the event characteristics model; and

a demand forecasting process that calculates a variation in load on the IT resources based on the predicted values of the variation over time and the number of the users moving that are calculated by the event influence calculation process.