Performance of SOA for Information Retrieval System Using Web Services

Information retrieval systems using web services are increasingly being used on larger databases and by more users. Current system allow users to connect to a multiple database either locally or perhaps on another machine. The resource demands limit the performance of IR systems especially as the size of text collections and the number of applications increase. Service oriented architecture (SOA) computing offers a solution to these problems. Systems based on distributed architectures can use resources more efficiently by spreading work across a network of workstations and by enabling parallel computation. An IR system is an ideal application to distribute across a network of workstations. The amount of information available and the number of people accessing data over networks is rapidly increasing. To meet future demands, a SOA based IR system must provide concurrent, efficient access to multiple databases collections located on remote locations. However, due to the mix of I/O and CPU intensive operations, IR systems present unique problems for distributed system designers. The disparity between I/O and processor speeds exacerbates these problems. Another concern is network traffic, since the amount of data transferred over the network by a SOA based IR system fluctuates considerably.