160 likes | 342 Views
A Study of the Impact of Compression and Binary Encoding on SOAP. Presented by: Alex Ng AWSA 2005 29 March 2005 Brisbane, Australia. Shiping Chen CSIRO ICT Centre. Paul Greenfield CSIRO ICT Centre. Alex Ng Department of Computing Macquarie University. Outline. Background SOAP
E N D
A Study of the Impact of Compression and Binary Encoding on SOAP Presented by: Alex Ng AWSA 2005 29 March 2005 Brisbane, Australia Shiping Chen CSIRO ICT Centre Paul Greenfield CSIRO ICT Centre Alex Ng Department of Computing Macquarie University
Outline • Background • SOAP • Binary XML • XML Compression • Experimental design • Result discussion
Background • SOAP Web Services has been very successful in platform-independent application • Performance Characteristics of SOAP • Typical SOAP message size is significantly larger than binary format • SOAP performance is affected by implementation and the choice of encoding style • SOAP itself is not processor intensive • Concerns: • The verbosity of text-based XML, processor and bandwidth requirements • The use of base64 encoding for binary data
Some Approaches in Optimizing SOAP • SOAP Message Transmission Optimization Mechanism (MTOM) • Recent W3C standard • XML InfoSet, WS-Security & WS-* compatible • Streams binary data as MIME message • Uses XOP to selectively encode individual SOAP elements • Binary XML • W3C XML Binary Characterization WG • Sun Fast Web Services • XML Compression • Schema Based: XMill & Millau • Dictionary Based: gZip
Research Questions • How effective is compression as a way of improving the performance of SOAP Web Services? • How effective are the proposals for binary encoding for XML likely to be improving the performance of SOAP Web Services?
Test Environment • Hardware: • 3GHz, • 1Gb memory, • 100Mhz Ethernet, • CeNTIE 1Gps WAN • Software: • W2003 Server Standard • ASP.NET 1.1 • Test Setup: • SOAP: ASP.NET Doc/Lic • Binary: .NET Remoting HTTP/Binary • gZip: SharpZipLib.dll • MTOM: translate our binary messages into MTOM format by hand
Latency Summary (lightly loaded network) • SOAP performs as good as other binary-encoded alternatives in test messages with simple structure and small number of XML tags • SOAP is not far behind in other cases. • The number of XML tags and structures affect the performance of SOAP • The overhead of Base64 encoding is not significant (in latency) • Compression always slow in our situations
Processor Time Per Request Compression Time not high in absolute value
Simple Medium Predicted Latency (ms) Predicted Latency (ms) The Impacts of Compression : Simple & Medium Bandwidth Bandwidth SOAP SOAP gZip gZip Remoting Remoting 100Mbit/Sec 100Mbit/Sec 16 7 26 15 8 6 10Mbit/Sec 10Mbit/Sec 28 13 30 17 13 7 1Mbit/Sec 1Mbit/Sec 31 147 72 37 67 20 500Kbit/Sec 500Kbit/Sec 279 51 118 58 34 127
Complex Predicted Latency (ms) The Impacts of Compression : Complex Bandwidth SOAP gZip Remoting 100Mbit/Sec 29 46 11 10Mbit/Sec 62 55 20 1Mbit/Sec 386 109 115 500Kbit/Sec 745 169 220
Conclusion • Both compression and binary encoding are effective at reducing message size • Compression works best with complex and repetitive structure • MTOM is effective with large binary attachment but not for message with small number of binary elements • due to the overheads of MIME and XOP elements • Compression appears to be useful for all • Exception: the least compressible messages and the fastest network