XML's overhead will open wallets?

re: XML's overhead will open wallets?

damien morton — Wed, 29 Dec 2004 00:18:00 GMT

For me, binary xml is way overdue. Im working on real-time financial applications, and have found that serialization and deserialization of xml messages is a very significant overhead.

I know that xml is text, and therefore 'open', but I dont believe that would be a problem if binary xml was standardised. As noted in the text of the paper, most "view source" commands are actually displaying a serialised version of their internal data structures. A standardised binary xml would have a plethora of "view source" tools available for it.

re: XML's overhead will open wallets?

Stephane Rodriguez — Wed, 29 Dec 2004 00:40:00 GMT

What about measuring tools? (http://www.arstdesign.com/articles/xmloptimization.html)

re: XML's overhead will open wallets?

Stephane Rodriguez — Wed, 29 Dec 2004 00:43:00 GMT

By the way, I think those pointing Xml overhead are very right when it comes to using taxonomies rather than simple xml.
As a corollary, what about a "simple xml" proposal (without entities, and I'll go as far as pointing without support for attributes either) ?

re: XML's overhead will open wallets?

Mike Champion — Wed, 29 Dec 2004 01:13:00 GMT

Stepahnie Rodriguez: Thanks for the XML Optimization link! That is the kind of "work smarter" thing I was talking about.

damien morton: I agree; the paper http://www2003.org/cdrom/papers/alternate/P872/p872-kohlhoff.html I cited in the XML 2004 stuff opened my eyes on that. The question is whether a standard binary XML format that meets the needs of the financial industry will also meet the needs of others, e.g. the wireless folks. That's an open question, but from what I've seen, I'm not terribly optimistic. I hope I'm unduly pessimistic!

re: XML's overhead will open wallets?

Kurt Cagle — Wed, 29 Dec 2004 13:16:00 GMT

One of the central problems with any new technology (not that XML is all that "new" anymore) comes from learning the best methodologies involved in the handling of that technology. Binary serialization of XML misses the point that XML is an abstract representation of data, and the moment that you move into the binary realm the abstraction goes away; at which point you're basically left with the whole DCOM/CORBA mess of attempting to rectify simplify data types, not to mention complex object types with differing methods of platform serialization.

Thus, certain sectors that are dealing with the fairly high bandwidth cost of that XML need to make some ugly decisions: Forego standardization and platform neutrality by choosing a specific format (albeit one not standards "blessed") for binary encapsulation of the XML, or rethinking the proposition concerning how much XML needs to actual go over the wire and be serialized in the first place. My suspicion is that better designed schemas, the use of "LOD" instances which can load in additional information as required, and other such optimization schemes could go a long way to alleviating the problems that XML faces in that space.

re: XML's overhead will open wallets?

Softwaremaker — Wed, 29 Dec 2004 15:38:00 GMT

+ 1, Mike

On a slightly different perspective and from a designing front, I have been advocating against the use of XML in all layers of a enterprise application esp. when tightly bound object technology is much more desired. In my presentations on SO(A), I have always preach on using service-messaging as communication b/w applications, NOT between tiers of an application.

However, many businesses are using XML Services as a communication mechanism JUST SO they can be seen as implementing an SO(A)... and of course, all for the wrong reasons.

Hence, many of them complain when performance suffers. What do they expect when they are making verbose calls to their own Data Access Layer via SOAP ?

re: XML's overhead will open wallets?

Jack — Wed, 29 Dec 2004 15:54:00 GMT

Both the links to your paper and presentations are broken. Could you please share the latest links ? Thanks.

re: XML's overhead will open wallets?

Mike Champion — Wed, 29 Dec 2004 17:39:00 GMT

Jack: Thanks! There was a problem in the PPT link in that it had some escaped spaces at the end. That's fixed. The other link worked ... It's possible that these are passoword protected and I just have a cookie set, but I've checked this on a couple of machines and a couple of browsers.

re: XML's overhead will open wallets?

damien morton — Fri, 31 Dec 2004 22:09:00 GMT

Kurt - I dont understand how moving from text to binary changes the level of abstraction.

Yes, you will need to come up with e.g. standardised representations of floats and ints, but I think that problem is far simpler than problem of managing text encodings in xml.

I also agree that any given binary xml standard cant solve everyones problems, but I think that any standard is better than no standard, if it makes for significantly faster processing and significantly more compact representations. From what I understand, those two goals are completely complementary. The Sun presentation on Fast WebServices is particularily interesting, in that they achieve a 5-10 times increase in processing performance, and a message size one fifth that of the equivalent soap message.
http://java.sun.com/developer/technicalArticles/WebServices/fastWS/

As far as LOD goes - for my problmes we already have LOD implemented, trimming down the message elements to the minimum possible, and still the xml processing overhead of our client-side application is too high - much higher than if the messages were binary encoded. Even the simple conversion of text to floating point accounts for 5-10% of our application performance.

Think about it this way: what is gained from converting an IEEE 8-byte double into a 12 digit text representation wrapped in ~10 character xml start and end element tags (all of which in various unicode encodings) as it travels between two parties?

We could come up with our own binary encoding, but we work in a heterogenous technology environment, and it would be better to use standardised libraries for a standardised encoding.

Today and tomorrow: Will Indigo heal...? Explicitness is good, especially for Contracts

Christian Weyer: Smells like service spirit — Thu, 10 Feb 2005 13:45:00 GMT

mikechampion s weblog XML s overhead will open wallets | Paid Surveys

mikechampion s weblog XML s overhead will open wallets | Paid Surveys — Sat, 30 May 2009 03:01:29 GMT

PingBack from http://paidsurveyshub.info/story.php?title=mikechampion-s-weblog-xml-s-overhead-will-open-wallets

mikechampion s weblog XML s overhead will open wallets | Uniform Stores

mikechampion s weblog XML s overhead will open wallets | Uniform Stores — Wed, 03 Jun 2009 13:31:42 GMT

PingBack from http://uniformstores.info/story.php?id=44251