EXI Comparisions EXI Emerging W 3 C standard

EXI Comparisions

EXI Emerging W 3 C standard, now in “final call” status on the standards track http: //www. w 3. org/XML/EXI/ Provides a more efficient, alternate way to encode the XML infoset. • Relaxes XML requirement for text-only • Can use schema information (if present) to encode XML document data more efficiently • Can also work (less optimally) without a schema

EXI & Military This has big implications for the military • We want XML for data interoperability • But XML is too verbose to use on many military networks • Poor string databinding characteristics push implementers to use custom binary protocols • Which results in silos of exellence, and nobody able to use the anyone else’s data without a lot of custom programming

EXI Measures of Effectiveness: • Document size • Databinding performance The first is obvious: we want to use XML everywhere, but the documents can be too big. EXI can make documents smaller, while being exactly equivalent to the original XML The second less so: we want to be able to efficiently get data from the document to a programming language, such as Java, C++, Python, etc. The second is a hard problem to measure; all sorts of measures of effectiveness

XML & Gzip Why not just gzip XML? • Doesn’t solve databinding issue at all • For many XML documents with numeric data & a schema, EXI is more compact

Size Comparison: Single DIS PDU IEEE DIS 144 100% XML 1387 963% Gzipped XML 545 378% EXI (No schema) EXI (Schema) 752 522% 115 80% the size of original binary!

Size Comparison • The Gzipped data has the inherent problems of databinding string data • The IEEE binary format is understood only by DISsavvy applications • The EXI schema-informed format is smaller than the original binary IEEE format, but at the cost of somewhat higher parsing complexity • The schema-less EXI format is fairly competitive with gzip (results vary depending on the nature of the data)