Browse Prior Art Database

%cc% Optimized Tenant Resource Allocation in SaaS Scenario Disclosure Number: IPCOM000198077D
Publication Date: 2010-Jul-26
Document File: 2 page(s) / 111K

Publishing Venue

The Prior Art Database


Background: In SaaS operation environment, there are large number of services and tenants. Tenants will subscribe one or more services. Every subscription requests a certain number of system resource (such as CPU, disk, memory) that claimed in SLA. These requests have some features: • A large number of requests will come in short time • Each request include different type of resource SLA • Some resource allocation request will take much longer time than others • Requests need to be processed in specified response time Problem: Usually operator allocate system resource manually and ad-hoc. It will cause: • System resource allocation is not balance • System performance would be lower • Global optimization algorithm is not fit for this scenario Introduction of this invention: It's an apparatus for tenant computing resource allocation in SaaS environment. This apparatus is targets on optimize the resource allocation to contain more tenants. The main idea of this apparatus is: • Classify SaaS tenants on-boarding request by their profiles, SLA and other information • Order tenants requests by SLA and tenant behavior estimation • Find and allocate some computing resources for new coming tenants under some optimize condition

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 86% of the total text.

Page 1 of 2

%cc% Optimized Tenant Resource Allocation in SaaS Scenario


Detailed components information:

    Request Resource Profile Pattern Identifier
- Identify resource profile pattern by data mining in history subscription database.


And create a resource profile for the request

    System Resource Monitor
- Calculate free and available system resource of each server Request Resource Profile Evaluator
- Evaluate each request resource profile by SLA and pattern
- Evaluate the resource profile of each server by monitor
Request Dispatcher
- Dispatch a request to resource allocation queue of one server. This algorithm include these steps:

- Choose all the request in a period of time. E.g. in one hour

A Resource Profile include:

    - Requested resource type
- Requested resource quantity
- Resource change trend
Pattern Recognizer
- Recognizer matches requests to the existed pattern by:
- Requested resource type and quantity
- Tenant history behavior
- Service resource feature


[This page contains 1 picture or other non-text object]

Page 2 of 2

- Set a main optimization target of Dispatcher. E.g.
- Min number of server
- Load balance
- Min response time

- Calculate resource allocation by global optimization algorithm
- Regard the resource profile pattern as the constraints
- If current resource request is satisfied, optimization

algorithm will consider resource change trend and refine the dispatching result.
Server Request Queue
- Store all the requests of each server Resource Allocator