XmlPerformance

XmlPerformance is getting attention due to their dominance of the network bandwidth. See http://www.xmlmania.com/news_article_1674-ZapThink-XML-Inefficiency-Choking-Networks-Market-for-High-Performance-and-Appliance-Approaches-for-.php

I hope there will be resources here later to discuss how XmlPerformance can be improved, and relative merits of alternate approaches.

---

Standards aiming at improving performance

In January 2005, the W3C ratified SoapProtocol V1.2 based "MessageTransmissionOptimizationMechanism" (MTOM), which uses Xml-binary Optimized Packaging (XOP) and Resource Representation Soap Header Block (RRSHB)to allow for efficient binary data transfer. See http://www.w3.org/TR/2005/REC-soap12-mtom-20050125/

• A Gartner Inc set of slides on this can be found at http://www.w3.org/2005/Talks/0420-yl-gartnerWS/.
• "Recommendations Support Faster Binary Data Transmission" at http://xml.coverpages.org/ni2005-01-25-a.html

---

Resources

Fast InfoSet and other implementations at http://builder.com.com/5100-6371_14-5534881.html?part=rss&tag=feed&subj=bldr

RoadMap to high performance XML by "Daniel Cazzulino at http://weblogs.asp.net/cazzu/archive/2004/07/09/HiPerfXML.aspx

---

Q: What is the performance of XML when compared to other methods?

A1: What do you want to measure? XML vs a text file with no markup? XML vs a text file with a proprietary markup? XML vs binary?

A2: Sans a comparison of XML vs non-XML, some numbers from "Benchmarking XML Parsers" (http://www.xml.com/lpt/a/Benchmark/article.html) by Clark Cooper may be illustrative. He tested six parsers in four different languages. His machine is a PII laptop with a fair but not extraordinary amount of processing power (232.65 BogoMips). CPU seconds to process a 1,264,240 byte file, 33% of which was XML markups (as opposed to content) were:

 C-Expat     0.380
 C-Rxp       0.740
 Java-xp     4.770
 Java-xml4j  6.770
 Perl        8.410
 Python     12.183

Which XML parsing package are you using in Python? There are several, and the speed varies widely.

---

Network throughput and processing speeds are quite different. Obviously XML is incredibly redundant and a prime candidate for compression, approaching a 90% or more compression rate.

---

CategoryXml