You are right. It is great technology. I have written a white paper on DB2 HADR and Snapshot technology. You can find this paper on one of the following Web sites http://www.docstoc.com/docs/4759241 or http://www.redbooks.ibm.com/abstracts/redp4278.html
As I stated, we use our own proprietary operating system to manage data on the disks. Our operating system has clear picture of all the data blocks all the time. When we take Snapshot in other words take backup, we only create a copy of data block headers. We don't touch data block at all. As you know Block header or inode consists just block address that is very little data. Hence inode copy can be created in matter of seconds without requiring a physical data copy operation. At the time of restore same thing applies, you are not restoring whole data set, only inode information is restored. The active filesystem starts pointing to the restore inodes. Snapshot technology is very interesting, it not only allows you backup/restore in matter seconds but also allows you to create a second copy of the database that you can use as independent database. The second copy is called clone. If you are more interested about this technology please visit www.netapp.com. You can find lots of collateral around DB2 I have co-authored on this Web site or IBM's (www.redbooks.ibm.com).
I hope this is helpful.
Please let me know if you have any question.
Snapshot feature you viewed in demo is functionality of NetApp's proprietary operating system called Data ONTAP. NetApp offers wide range of storage products starting from few thousand to hundreds of thousands dollars. If you look from DB2 customer point of view. If a customer has NetApp storage and want to deploy DB2 on it, the Snapshot comes almost free. I said almost because you many need to pay some fee for DB2.
Please let me know if you have any question or I can help you in anyway.