Soap performance enhancement for high volume messaging / Ali Baba Dauda

Ali Baba , Dauda (2018) Soap performance enhancement for high volume messaging / Ali Baba Dauda. Masters thesis, University of Malaya.

	PDF (The Candidate's Agreement) Restricted to Repository staff only Download (200Kb)
Preview	PDF (Thesis M.A) Download (1493Kb) | Preview

Abstract

The emergence of high-volume data exchange like business-to-business and computational sciences that are mission critical and always persist over time have exacted distributed systems and applications to be fast. SOAP is one of the best protocols using XML to exchange message but the XML is too verbose and slows the communication process. To this end, message exchange accumulates overhead and high response time resulting to slow communication and message lost during the transmission. Therefore, reducing the response time and overhead will enhance the communication process. To achieve this aim, LZ77 compression algorithm is modified to encode more symbols. The algorithm is then integrated into two Web services with HTTP and JMS bindings. The HTTP Web service as the benchmark and the JMS as the prototype Web service. For both Web services, the server holds provider, compressor and controller classes and the client contain consumer and decompressor classes. The client invokes the server to establish WSDL contract and communicate via the relevant protocol. Two messages formats, normal and compressed (modified algorithm) with the size ranging 1MB - 22 MB were generated and executed 50 times in both web services. The performance effects of the message formats for the Web services were recorded. The metrics of the Web services used are the payload overhead, server response time, client response time and compression/decompression overhead. The payload overhead, server response time and compression overhead were analyzed at the server side. While client response time and decompression overhead were analyzed at the client side. Average values of the metrics were calculated and the transaction response time is obtained as the sum of response times and the overheads at both endpoints. The metrics were plotted against the message sizes and the effects were analyzed. The findings demonstrated that the compressed JMS binding on SOAP messages recorded low response time and low overhead compared to the compressed HTTP binding. In the compressed HTTP binding, the internal process at the client side regularly claims memory while creating available space for incoming messages resulted in producing of spikes leading to high overhead. Out of the 50 executions for 12 transactions, compressed HTTP binding delivery failed 6 times, and compressed JMS binding failed 2 times. While for the normal HTTP binding delivery failed 5 times, and normal JMS binding failed 2 times. The overall findings observed that with the modified LZ77 algorithm, SOAP over JMS has proven to be better than the SOAP over HTTP. The SOAP (with the modified compression algorithm) over JMS is a good technique for exchanging high volume messages when low response time and guarantee of delivery are needed in the communication.

Actions (For repository staff only : Login required)