Service-oriented Data Denormalization for Mediawiki

From Master Projects
Revision as of 11:43, 17 September 2010 by Gpierre (talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Service-oriented Data Denormalization for Mediawiki
status: finished
Master: project within::Multimedia
Student name: student name::Peng Li
Start start date:=2010/01/01
End end date:=2010/07/01
Supervisor: Guillaume Pierre
Poster: has poster::Media:Media:Posternaam.pdf

Signature supervisor



Many techniques have been proposed to scale web applications. However, the data interdependencies between the database queries and transactions issued by the applications limit their efficiency. To overcome the limitation, service-oriented data denormalization technique restructures a centralized database into several independent data services, and reorganizes the application into a service-oriented architecture. Each data service handles specific business logic, and has exclusive access to its own database. This project aims at applying service-oriented data denormalization technique to MediaWiki. First of all, I have split MediaWiki‟s data into many independent data services by studying the data access pattern in the code of MediaWiki extensively. Secondly, I have defined and developed these data services in terms of Web services in PHP. Each Web service has its own database, and exchanges data via HTTP and JSON (JavaScript Object Notation). After denormalizing MediaWiki, I have scaled individual Web service by applying special techniques on each Web service according to its own characteristics. Finally, I have studied the scalability of the denormalized MediaWiki, and demonstrated that the maximum sustainable throughput grows linearly with the number of hosting servers used.

Final thesis