Do we need an alternative to human-readable text XML? Is the angle-bracketed payload of text XML becoming too much for our systems to bear? In a recent post, I reported on some swipes at efforts to introduce Binary XML.
Microsoft's Rich Turner has issued a rebuttal (posted here) to the Binary XML naysayers. He states, in no uncertain terms, that Binary XML is here to stay, and we need it. "What I take greatest exception to is the notion that we don't need a BinaryXML standard. Whether or not the W3C's BinaryXML work is the best approach to the problem, the problem exists - parsing text-XML is complex and time-consuming. There is a very clear need to alleviate the XML performance problem - so much so that IBM recently bought DataPower Technology Inc - a company that sells XML processing devices."
Rich also observes that "the cost of parsing of text-XML completely overshadows the cost of transporting of the resultant angle-bracketed payload. If we, the IT industry, can work together to develop a standard approach to encoding the InfoSet in a manner that simplifies the parsing and processing of XML and simultaneously decrease the size of the payload itself, then we'll have created an alternative encoding of XML. Note that this would be an ALTERNATIVE encoding, not a replacement for text-XML."
Binary XML will also bring XML standardization to devices which cannot support text encoding, such as embedded controllers, limited bandwidth and satellite communications, Rich adds. "By creating a BinaryXML standard, we'd see XML adoption in scenarios where its use is not currently widespread and we'd see applications and systems in those spaces increasingly be able to communicate with one another far more freely than they do today. This would benefit us all."