Identifiers and Handles

Desirable properties of identifiers

The following are desirable properties of identifiers in the digital library:
  • Location independent name
  • Globally unique names
  • Names that are persistent across time
  • Choice of automatic generation or use of assigned names

  • Fast resolution of identifiers
  • Replication and caching of objects and resolvers
  • Decentralized administration
  • Change control

  • Support from standard user interfaces
    • A handle is an identifier that satisfies the first group of criteria. The CNRI Handle Management System is an implementation of the architecture that satisfies the middle group of criteria. Current activities are aimed at having handles incorporated within standard user interfaces.

    Syntax

    The syntax of a handle is either:

    naming_authority / locally_unique_string

    or, for use in Internet applications,

    hdl:// naming_authority / locally_unique_string

    Examples

    cnri-1/1995.02.12.16.42.21;9 (date-time stamp)
    berkeley.cs/ucb-cstr-94.45 (semantic name)

    Naming authorities

    Names of naming authorities are unique for all time.

    The Handle Management System

    The Handle Management System has three components:

    Handle Generators

  • Create globally unambiguous handles that can be associated with digital objects.

    • Handle Servers

  • Store the handle and associated location information.
  • Process requests from clients to resolve handles.
  • Return location data to the client.

    • Handle Server Directory

  • Provides a list of Handle Servers.
    • The Handle Management System is in operation and available for public use.

    Resolving a handle

    The following diagram shows the most common way that the handle server can be used to resolve a handle and return references to the objects that it identifies.

    A hash table is used to indicate which of the server computers holds the information associated with a specific handle. There are three standard configurations.

  • The client can connect to a caching server, which knows the hashing algorithm, has a copy of the hash table and holds a cache of recently used handles. (This is the configuration shown in the diagram.)
  • The client program can know the hash algorithm and hold a copy of the hash table. It may maintain its own cache. This is equivalent to the client acting as its own caching server.
  • The client can connect directly to any handle server. The server will do the hash table look up and pass the request to the appropriate server computer.

    • Handle Servers return a choice of all or subset of stored data.

    Data associated with handles

    Handles and the Handle Management System were designed for use in the digital library, but they can be used to store a wide range of data:

  • Object references to Repositories
  • Universal Resource Locators (URLs)
  • Electronic mail addresses
  • Public keys
  • etc.
    • wya
    February 1, 1995     ÿ