Browse Prior Art Database

A Mechanism & Apparatus for Multi-tenant Job Processing

IP.com Disclosure Number: IPCOM000198981D
Publication Date: 2010-Aug-19
Document File: 3 page(s) / 81K

Publishing Venue

The IP.com Prior Art Database

Abstract

Many ISVs are thinking of moving their applications to SaaS (Software as a Service) market. Some of these applications will leverage backend jobs to handles some maintenance or summary jobs, for example to summarize the average page view numbers in some selected periods. Job execution time & resource consumption estimation is essential input for a effective Job schedule system. In MT scenario, the same job may have different job execution time & resource cost for different tenant. So It's difficult to accurately estimate the job execution time & resource consumption for each tenant.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 3

A Mechanism & Apparatus for Multi-tenant Job Processing

1) Background & Problem
Job execution time & resource consumption estimation is essential input for a effective Job schedule system.

In MT scenario, the same job may have different job execution time & resource cost for different tenant. So It's difficult to accurately estimate the job execution time & resource consumption for each tenant.

Problem: How to estimate the job execution time & resource consumption in multi-tenant job processing system

Input:
* Job info
* Job start & finish time requirement
* Job execution time & resource consumption estimation

Job Scheduler

 MT Job Execution System

2) Solution Main Idea

Classify tenants into several group based on tenant's static info, e.g. user number, industry, subscribed job list etc.

Generate job estimation model based on tenant group dynamic info and job execution history. To find out the dynamic info and execution history fitting curve

Generate estimation result for new jobs & tenant using estimation model and tenant's current dynamic info

T1 T2 T3

T4 T6

T21 T22

Job list , including Job estimated execution time and resource consumption etc.

Tenant Group Classifier

1

MT Job Estimation Module Generator

MT Job Estimation Result Generator

Job Scheduler

2 3

Tenant dynamic status monitor

 MT Job Execution System

MT Meta Repository

(Tenant Characters, Job execution history, tenant dynamic prams monitor result)

Job execution result monitor

3) Overall Process

1

Page 2 of 3

Classify tenants into several tenant group

  Gather tenant characters/static info

 Gather tenant dynamic info & job execution result

Generate estimation model for each job in each tenant group

 For the job list, identify the tenant identity for each job

Classify the tenant into a tenant group

Get the tenant group's job estimation model

Generate estimation result for the job

4) Details
a) The tenant group classifer will :
- Classify tenant into groups based on tenant static information (characters) User number, industry, subscribed job list etc.
- Classify based on tenant similarity: One possible similarity definition is that : If 80 percent of the property in tenant static info are identical , then tenants are similar. For each property, if it's continuous then identical means their variance is in certain scope.

b) The Job Estimation Model Generator will:
-Generate the mapping model of tenant dynamic status with job execution...