Sun released X4500 (Thumper) - a beautiful new 48 disk computer with 24 Terabytes of storage space with 2 dual-core AMD processors and 16GB of main memory. See also X4600 and Sun Blade 8000. While Gigabytes may not be new to people who have bought an iPod or a computer system, it perhaps gives a better sense of capacity and size if expressed in other units.
So here are my calculations:
One fully loaded Thumper is about 50 years of continuous mp3 music! - more than what you can hear in a life time; or one year worth of playing continuous DVD quality movies! or photo IDs of an entire country like India with one billion people (children included)
If you recorded your voice everyday in telephone quality sound, you, your grand children, great grand children .. also did that, then you need 8 centuries to fill this box! However if all the international calls to or from US are wire tapped and recorded, then it takes about 2 days to fill the box! (per FCC report, 80 billion minutes were spent by people over international calls in 2004) To wiretap every telephonic conversations in a year, you would need about 18 racks each fully loaded with 10 Thumpers. Not an expensive deal for the government!
Imagine a fully loaded rack which can take 10 Thumpers, then it means every thing above gets multiplied by ten!
If you are wondering if there is a filesystem that can handle this capacity and manage the storage for years without having to worry much about disk failures, take a look at ZFS - the last word in filesystems.
If you have other creative ideas on how you can use this capacity on a single system, please do share.
Tag: OpenSolaris

dbmopen(%D,$ARGV[0],0444);for(keys %D){print "$_ $D{$_}\n"}
Scripting languages like PERL make it really easy to write "your own", rather than try to find "some other tool" that does the work. It is perhaps faster to write your own.I have been using DBM (see dbm(3UCB)) databases for all my database needs. Operating on DBM databases in perl is extremely easy. They are quite fast when you do not really need a full fledged relational database. Being a bit allergic to database classes in school, I had avoided using big relational databases like Oracle or MySQL.
When I started writing a tool which gathered huge amounts of data from various sources and then cross-checked and audited them, DBM files turned out to be quite inadequate. I desperately needed a relational database.
I tried Apache Derby, a pure java implementation of a relational database. Wow! it just worked out of the box on Solaris. To start the server just run the startNetworkServer.ksh after setting the $DERBY_INSTALL path in that script. Derby can be used either in server mode or embedded in the application just like DBM databases. Derby site also has a handy and excellent SQL reference guide.
What more, netbeans has a really neat GUI way to get into the database. You could also point OpenOffice to the derby jar file, and can connect to Derby database with URL like jdbc:derby://localhost:1527/mydb1. There is also a CLI tool org.apache.derby.tools.ij.