First NITF, Now NewsML
by Christopher J. Feola
It may be a measure of how much our industry has changed that the following sentence could come from the mouth of an NAA vice president:
"News Markup Language is a boxing structure that can contain multiple pieces, and can relate those pieces to other pieces."
A boxing structure? Gee, remember when PostScript was the cutting edge of technology?
Get over it. Those days are gone, as John Iobst, NAA vice president for production operations and research, would be the first to tell you. Nowadays, newspaper companies have to help develop things such as the News Markup Language. NewsML is an eXtensible Markup Language document type definition for news. That’s right, NewsML is an XML DTD. XML is a metalanguage–a way of describing data–and a DTD is basically an instruction manual for an XML document.
Now those among you who pay attention to such stuff are probably asking, "Wait a dang minute! Didn’t we just do this with the News Industry Text Format? Do we have to do all that again?"
Two answers: Yes we did, and no we don’t. Here’s Iobst to explain: NITF–approved last year by The International Press Telecommunications Council (TechNews, May/June 1999, p. 22)–"is a text format. It really is not a multimedia format. It’s a story format."
NewsML actually builds on NITF and other existing formats, such as JPEG for images and MPEG for video. Each type of content uses its own format, and NewsML ties them together. "Think of it as a shipping label and packing list for other file types, like NITF, other XML formats, MPEG-7, etc.," says Glen Cruickshank, a KPMG senior manager and member of the committee that wrote NITF. "While those files can be embedded inside a larger NewsML file, they can be linked as well, and most likely should be."
That’s what Iobst means by a boxing structure. Let’s review: "[NewsML] can contain multiple pieces, and can relate those pieces to other pieces in the package and elsewhere."
Why is that important? "NewsML allows for complements and alternates," Iobst says. "Alternates could be multiple versions of [the same] story, or TIFF and JPEG versions of [the same] photo. An AP European wire might carry the same story in three languages."
Complements, on the other hand, are things that go together. "You could have a story, two sidebars, three photos, etc.," says Iobst. "And of course, you could have alternates for the complements."
NewsML currently is in beta. (NITF already is supported by AP’s new XML feed and some editorial-system suppliers.) Some final adjustments are being made, and test implementations are in the works at Reuters and Bridge News. "[IPTC] will have significant conversations at a meeting right before the IFRA 2000 Expo," Iobst says. An association of wire services and other news organizations, IPTC is hoping for final approval of NewsML during that October meeting.
The beta DTD for NewsML v1.0, together with a functional specification,
supporting documents and background papers, can be found on IPTC’s
Internet site at www.iptc.org/NewsML. The DTD is available as a rights-free
standard, but remains IPTC’s intellectual property. 
Feola is chief of technology at Belo Interactive. E-mail, chris.feola@belointeractive.com.
Web Traffic Outsourcers Deliver
by Andrew Bowser
Beset last year with burgeoning site traffic, washingtonpost.com Vice President of Technology Eric Schvimmer faced yet another expensive round of server upgrades.
 |
|
Digital Island's infrastructure pushes content closer. |
"The rate of buying was killing our budget," he says.
To staunch the rate of server growth, Schvimmer tapped Akamai (www.akamai. com), one of a new breed of content-delivery services specifically geared to high-volume World Wide Web sites eager to get content to users’ desktops quickly.
Akamai and other players, including Digital Island (www.digitalisland.com) and iBEAM Broadcasting Corp. (www.ibeam. com), use networks of geographically distributed servers, allowing them to serve up content to many users simultaneously and quickly–ideal for handling peak loads during big news events.
Washingtonpost.com beta tested the technology last summer and had it fully deployed by September, just in time for elections and hurricane season.
Both text and graphics are delivered from the Post’s own server, but the graphics are cached at an Akamai server closer to a particular user. After that, users located near that Akamai server get the cached graphical content. Yet the low-bandwidth portion still is delivered by washingtonpost.com, keeping Schvimmer and colleagues in control of page accounting.
The Post isn’t alone. WSJ.com used Digital Island to manage traffic spikes generated by a Super Bowl commercial; the U.K.’s Financial Times uses the provider not only for content delivery but also for hosting and country-specific ad targeting. Latimes.com used the services of Sandpiper, another content-delivery company that recently merged with Digital Island, to assist with Academy Awards coverage.
Once set up, the service is a seamless and essentially invisible part of day-to-day operations. Elements washingtonpost. com wants to serve out simply are assigned an Akamai URL before being published. The site’s templated content is set up to automatically reference the content-delivery provider.
Schvimmer says the monthly service charge is justified by the tradeoff
in what he would be paying for licensing, depreciation and total-cost-of-ownership
of additional servers. "Even though there was a premium, I determined
that did not exceed the cost of building and running it myself,"
he says.
Bowser is a Brooklyn, N.Y., free-lancer. E-mail, andrew@bowser.com.