The agency brings a galaxy of Web sites into the fold
- By William Jackson
- Jul 02, 2003
On Jan. 31, a server in the basement of NASA headquarters in Washington operated the NASA.gov
Web site with a single shared T1 connection to the Internet.
At 11:45 p.m. Eastern time that night, all the traffic cut over to a brand-new portal hosted in an AT&T Corp. data center.
'I remember it well,' NASA spokesman Brian Dunbar. 'The next day was when we lost the Columbia.'
Within 20 minutes of the space shuttle Columbia's breakup during re-entry, traffic to NASA.gov spiked from 2 Mbps to 175 Mbps.
'You can't really call anything connected with that day a success,' Dunbar said, but 'people were able to get to the updates we were putting out on the Web.'
The high-bandwidth data center connection plus thousands of pages cached across AT&T's network kept information available to the public as well as news media.
Last redesigned in 1997, NASA.gov's presentation had been limited by the browsers of the time. Meanwhile, the site had evolved into a decentralized collection of pages.
'We had about 3,000 sites in the domain and about 4 million pages,' Dunbar said. 'There was no guarantee that from any NASA Web page you could get to any other page.'
Nor was there any central content management. During the last Hubble space telescope service mission, sites hosted by the Johnson, Goddard and Kennedy space centers and a research center all focused on different aspects of the mission.
'At one time we had three different launch dates listed,' Dunbar said.
To bring order to the web of sites and take advantage of higher bandwidth and modern browsers, NASA in September issued a request for proposals for a new portal. Four finalists received $100,000 each to refine their proposals.Done in phases
At the turn of the year, NASA contracted with eTouch Systems Corp. of Fremont, Calif., to oversee development and hosting of the new site and to provide content management.
Despite the size of the project, it wasn't particularly challenging, eTouch federal program manager David Valliere said, because his company didn't have to do the work all at once or all alone. 'We typically phase things,' Valliere said. 'We don't take big bites.'
To nibble away at the project, eTouch brought in subcontractors Speedera Networks Inc. of Santa Clara, Calif., for caching and content delivery; Sprint Corp. for site hosting in a pair of East and West Coast data centers; and Critical Mass Inc. of Calgary, Alberta, for the page design.
'The idea was consistent look and feel,' Valliere said. 'NASA has a very deep amount of content, and it tends to be technically driven.'
To manage content, the contractors had to come up with a consistent taxonomy describing data from a wide variety of disciplines and sites so that the search engine could list things with the proper relevancy.
'NASA had been aware of this and had been working on it for years,' Valliere said.
The new site uses a search engine from Verity Inc. of Sunnyvale, Calif.
When the new portal opened for business Jan. 31, AT&T housed it temporarily under an old contract. On May 1 it moved to Sprint data centers and the Speedera edge network for caching and delivery. Speedera caches the pages on servers in 20 countries, provides Domain Name System services, and tracks network and server status to route page requests to the nearest server.
The contractors are taking three approaches to incorporate various existing NASA sites into the new portal.
Some major sites will remain on their own servers with portal links and a redesign to match the look of the portal. Smaller sites could be left on their own servers but incorporated into eTouch's content management system. The portal itself can generate content for still other sites.
The top 20 NASA sites will come into the fold first, and about 300 sites should be incorporated by the end of eTouch's contract Sept. 30. NASA headquarters information and links to some other information already are up on the portal page.
'At this point, we haven't incorporated any other sites wholesale,' Dunbar said.
A site for the Mars mission, redesigned to match the new portal, is slated to be among the first sites incorporated.
And the agency last month launched My NASA, a personalization tool. Content will be tagged with metadata to associate it with one of nine channels, such as astronomy or aeronautics. Users can specify channels to direct information to a personalized page.
'We probably aren't as far along as we would have liked, but we're learning,' Dunbar said.
NASA is a distributed organization, and its sites will continue to reflect that. The space agency wants to save money by consolidating pages on fewer servers, but there will always be multiple servers handling its material.
'We're not just taking 3,000 sites and saying, 'You're in the portal, shut your server down,' because the content won't allow that,' Dunbar said.
The publishing process will be consolidated, however, and site management will be more centralized.
NASA is examining millions of legacy pages to see whether they belong in the new system. Some will be scrubbed; many more will have to adapt to the new navigation system.
The agency will consider other ways of taking full advantage of its new bandwidth, such as holding more conferences and NASA TV webcasts.
'We're interested in getting some solid data on the upper limits of audience interest in those kinds of things,' Dunbar said.
William Jackson is freelance writer and the author of the CyberEye blog.