Scalable Web Hosting. Cost-aware resource provisioning in Azure

From Master Projects
Jump to: navigation, search

has title::Scalable Web Hosting. Cost-aware resource provisioning in Azure
status: finished
Master: project within::Internet and Web Technology
Student name: student name::Radu Nitu
Start start date:=2013/05/15
End end date:=2014/11/15
Supervisor: Thilo Kielmann
Second reader: has second reader::Stefania Costache
Company: has company::Amsterdam Software
Thesis: has thesis::Media:Thesis.pdf
Poster: has poster::Media:Posternaam.pdf

Signature supervisor



The research community as well as the industry area is more and more interested in the services provided by the cloud infrastructures. These services come with different price categories and are paid depending on the usage time and the performance specifications of the rented resource. Thus optimizing the right number of resources required by an application is highly important from the customer point of view.

Companies are often interested in the web hosting capabilities of the cloud infrastructure because it can offer automatic scaleout of web sites and background services and moreover it provides a better fault tolerant medium than the company itself could manage and support. The Azure cloud platform is an example of such an infrastructure that can provide the above mentioned services.

The cost effective rent of cloud services can be obtained with the help of a cost-aware resource provisioning algorithm. This should be able to determine the performance characteristics of the application and to minimize the renting price such that a certain level of service quality expressed in terms of Service Level Agreements (SLA) is maintained.

We are interested in the design and the implementation of such an algorithm that is compatible with the Azure cloud platform. This solution should ease the deployment of web applications and background worker applications to Azure, based on the application performance and usage characteristics, keeping to a minimum the renting price.

Additional information: The project work will be carried out at the company Amsterdam Software, under the supervision of Ries Vriend (CEO at Amsterdam Software) and Dr. Ing. Thilo Kielmann (VU University). The algorithm will be evaluated and tested with the company's main web product ¨FBO One¨ and the development process supposes the following steps:

define the SLA of the web application and other parameters needed by the cost-aware algorithm. extend the existing company tool (able to deploy web applications to a private hosting environment) to interact with Azure and to perform the cost-aware resource provisioning, plus management functionality. compare the company private hosting environment with the cloud hosting models and assess the optimal solution from various points of view (the operating effort, renting cost, performance, reliability/availability) analyze the possibility of offloading on demand instances from the private hosting environment into Azure, provisioning an optimal number of resources.