Let it flow: Building mashups with data processing pipelines

Authors: 
Börnstad, B; Pautasso, C
Author: 
Börnstad, B
Pautasso, C
Year: 
2009
Venue: 
Proc. Service-Oriented Computing-ICSOC 2007, LNCS 4907
URL: 
http://www.springerlink.com/index/xv7560067g11p427.pdf
Citations: 
0
Citations range: 
n/a

Mashups are a new kind of interactive Web application, built out of the composition of two or more existing Web service APIs and data sources. Whereas “pure” mashups are built relying entirely on the Web browser as a platform for integration at the presentation layer, this approach is not always feasible and part of the mashup integration logic must be deployed on the Web server instead. In the case study presented in this paper, we explore a novel approach to build mashups out of heterogeneous sources of streaming data. In particular, we introduce a layered mashup architecture, with a clear separation between the mashup user interface and the mashup integration logic run by a mashup engine. To do so, we show how to build a mashup application that displays in real time the location of visitors connecting to a Website. The integration logic feeding a map widget by processing the Web site logs is developed using a data flow model that connects a set of reusable and heterogeneous components into a data processing pipeline. We conduct a brief performance evaluation of the mashup showing that the pipeline introduces minimal delay and discuss several extensions of the mashup.

None
Login or register to tag items