A System for Persistent Names
By Catherine A. Rey
Managing content on a large Web site can be complicated, especially
if the quantity and complexity of that content grows, and the ways in which your customers retrieve and use that content evolves. At some point, you may need permanent identifiers that are independent of server location for a large collection of photographs, an inventory of merchandise with product codes, or a collection of documents that you'll eventually archive elsewhere. That's where the Handle System comes in.
The Handle System uses objects called handles to keep track of location and other information about data. I'm a Web-site manager, not a network architect, and I don't offer handles as a solution to all Web-site management problems. In fact, as part of the Handle System development group at the Corporation for National Research Initiatives (CNRI), I've proven that handles shouldn't be used for site navigation because most Web pages don't need to be permanent. For permanent content, though, handles are extremely useful.
At CNRI, we're often asked to compare the Handle System to DNS and other name services currently in use. Two simple points to make are that we believe the Handle System is more appropriate for large numbers of digital objects than DNS, and that the DNS administrative model argues against using it as a general-purpose name system. DNS administration typically requires a network administrator, and has no provision for administration per name by anyone other than a network administrator. (A longer discussion of DNS and more comparisons can be found online; see "
