[thechat] Sloan Digital Sky Survey

[rudy]

http://skyserver.sdss.org/

gorgeous, simply gorgeous

rudy


postscript for database afficio... afishio...  fans --

here's how i found out about sdss

i'm reading this database article, A Talk With Database Guru Jim Gray
(http://www.fawcette.com/interviews/gray/), and it's mentioned on the 3rd
page --

   SkyServer is like TerraServer but looking the other way. It has a 0.25
arcsecond picture of the northern sky (incomplete right now, but that is
where we are heading). In addition to the pixel data, it will have about 200
million galaxies, and about a million spectra. It is a data mining
challenge. So far we have been using pure SQL, but we are experimenting with
cubes and data mining. The project focus is to understand the structure of
the early universe, compute the cosmological constant to an extra decimal
place, and perhaps find out about dark stuff (the other 95 percent of the
stuff in the universe). Now that we have the SkyServer working, we are
converting it to a Web service, and working to federate it with the other
astronomy archives to convert the Internet to being the World Wide
Telescope. That's where I am putting most of my energies at this point. It
is a very good test of the Web services concepts.

remember, this is a microsoft guy talking -- does "convert the Internet to
being the World Wide Telescope" make any sense to you?

further down in the article --

   When data volumes get large, you can't grep them any more, you have to
use an index. That's when you need some kind of database. My group has been
trying to figure out how much you should put in the database-everything or
just the indices. As I get into it, I find it is a lot easier to put
everything in the database: photos, videos, images, and of course, text.
Sometimes it is just BLOb data, but often you want to "parse" the data and
extract as much meta information as possible. For example, we built a
database that has all our personal photos. The classic design would have
these photos in a file system and just the metadata in the database. But it
is so much easier to manage when everything is in the database. There is
only one thing to back up and to secure-everything is just simpler. There
are fewer design decisions to make. My prediction is that eventually all
storage systems will evolve to be database systems.

well, duh

of *course* the database vendors want that (it's been this way since
mainframe days when ibm was the only database vendor worthy of
the name) -- the more you store in the database, the more servers you
need...  think per-machine software licensing...

finally, this article has an unintentional bit of humour, that's actually on
every page but i didn't notice it until the last page, where gray does an
item by item analysis of sql/server -- quality, reliability,
manageability/usability, functionality, and performance -- as stacked up
against its competitors (any guess as to which product comes out on top?)

at the bottom of the page is a banner ad saying White Paper Sponsor-- Oracle