NASA's site carries star data
- By Bill Murray
- Mar 23, 1998
How do you serve up a terabyte of data that gets downloaded from the World Wide Web at
a rate of 1.7 million files per month?
The High Energy Astrophysics Science Archive Research Center at Goddard Space Flight
Center in Greenbelt, Md., relies on a 333-MHz Digital Equipment Corp. AlphaStation 600
Hypertext Transfer Protocol server with 384M of RAM and nine Web addresses.
The HTTP server has to be stable, because users worldwide want it available 24 hours a
day, said Phil Newman, HEASARC's webmaster. But the server has gone down occasionally,
sometimes because Goddard facility managers shut off the building's power for maintenance,
HEASARC uses Sybase Inc.'s SQL Server 11.0.2 relational database manager to
cross-correlate tables of astronomical data stored on optical jukeboxes and give pointers
to the data through the W3Browse archive browser.
The jukeboxes, which hold as much as 1.5 terabyes, include a Hewlett-Packard Co. 200T
Optical Disk Library storing 180G and an HP SureStore Optical 600fx jukebox storing 550G.
HEASARC bought the jukeboxes in 1993 and launched the Web site the following year. The
retrieval software is MagnaVault from Tracer Technologies Inc. of Gaithersburg, Md.
Visitors to the site at http://heasarc.gsfc.nasa.govmust
have Microsoft Internet Explorer 2.0, Netscape Navigator 3.0 or later versions of the
"We don't want to get too far ahead of the technology curve," White said.
"Many of our users are one or two versions behind" the latest 4.0 versions of
the Microsoft and Netscape browsers.
Users can retrieve data by File Transfer Protocol or through the W3Browse browser.
HEASARC, established in 1990, archives X-ray and gamma-ray astronomical data. Because
Earth's atmosphere absorbs most such rays, NASA and the European and Japanese space
agencies detect them through telescopes attached to satellites.
European and Japanese officials regard HEASARC as a good repository and willingly share
their X-ray and gamma-ray data, White said. The data streams in constantly, and a team of
astronomers, clerical staff and programmers turn it into usable form. Each day, HEASARC
adds as much as 3G to the Web site.
One big task is to make the data readable through the Flexible Image Transport System
(FITS), a raw-data file designed by astronomers, White said.
HEASARC also must transform old data sets, such as those stored in binary format, into
FITS so that users can retrieve the historical records. Unlike image or text files, FITS
files require X-ray tools to interpret them.
Unix still works best for HEASARC's needs, Newman said, despite the lower prices of
servers with Intel Corp. processors and Microsoft Windows NT.
"We are married to the Unix philosophy, but not necessarily to Digital Unix,"
Procurement is easier under NASA's indefinite-delivery, indefinite-quantity Scientific
and Engineering Workstation Procurements, Newman added.
"It's a paperwork nightmare if you have to go through anything other than SEWP
II," he said.
HEASARC will evaluate digital video disk storage and the Java language for the Web
site, White said, but the center probably will wait until they work better with different
Center officials will also continue making price-performance comparisons of managed
optical jukeboxes vs. disk farms, in light of the dramatic price drop in hard drives.
Managed optical jukeboxes cost about one-third as much as disk farms for the same
amount of storage, he said.