The Handle System: and its role in a Digital Object Architecture

1
The Handle System and its role in a Digital
Object Architecture
Robert E. Kahn CNRI
Workshop on Frontiers in Distributed Information
Systems Presidio of San Fransisco July 31
August 1, 2003
2
Objective of the Framework
Heterogeneous Networks
Information Systems
Information Systems
Networks
Seamless Whole
Internet objective Best-effort Packet
Delivery
Seamless Interoperability
Federating Heterogeneous Systems
3
Internet Comparison

• IP Addresses ? Machines
• Gateways (now routers) help with access
• TCP handles end-end issues
• Remove duplicate packets
• Restructure the arriving fragmented stream
• Perform end-end error detection retransmission
• Provide flow control

4
Further Scoping the Problem
Time to Resolve Query
Complexity of Query
Initial Focus on Queries with Complexity Zero
5
Global Identifier Interoperability
Cross-Industry Resolution
Literary Music Video Financial Grid Enum
RFID
Some Existing Approaches
SimpleLookup URL IPaddresses Unfederated
Databases
6
Basic Attributes of the Approach

• Digital Objects (i.e. Data Structures)
• Unique Identifiers ? Digital Objects
• Resolution Administration Mechanism
• Maintains Uniqueness of Ids ? DOs as long as they
  persist
• Maps Ids ? Useful State Information
• Is distributed and scaleable
• Does not involve complete search

7
Digital Object

• Set of elements, each of ltType, Valuegt
• Parsable across heterogeneous platforms
• One element must be the unique identifier
• Properties Record contains metadata
• Transaction Record records usage
• Most users wish to access its Essence
• Key Metadata is part of the Essence

8
Object-Oriented Programming
Digital Object Model
Disseminators
Methods
Internal Data Structure
Digital Object
Access to the object is subject to control by the
owner. For example, a market in disseminators is
possible.
The internal data structure is not directly
accessible by the programmer
9
Purposely Silent about

• What Types
• What Type of Types
• What Values
• What metadata or metadata schema
• What state information in Handle Records
• Policies and Procedures in general
• There are policies for Global, however

10
A Range of Possibilities

• Identifiers are persistent e.g. DOIs
• Identifiers are transient e.g. Grid
• Identifiers are resolvable
• Resolution information is not accessible
• Digital Objects are fixed, unchangeable
• Access to Digital Objects is fixed, even if DOs
  are changeable

11
Repository Notion
Logical External Interface
Any Hardware Software Configuration
RAP
12
Nesting of Repositories
Core
Structure
Content
Aggregation De-aggregation
Core Interface must be present at each
level Other levels could be separately defined
later
13
Federated Repositories

• Key issue is commonality of interests in
  accessing information from multiple repositories.
• Financial Information is prime applications area
• Metadata Registries allow for searching based on
  user-supplied inputs. The use of handles
  (however branded) can simplify access.
• Access via local repositories is an operational
  desirable capability.

14
MetaObjects Metadata Registries

• MetaObjects provide a structural basis for
  indirection and for organizing information
• Metadata is used to characterize digital objects,
  to access their identifiers and to assist in
  cross referencing
• Metadata Registries provide uniform access to
  metadata.

15
Handle Format
2304568.40/12345678
Naming Authority
Item ID (any format)
Suffix
Prefix
In use, a Handle is an opaque string.
Corporation For National Research Initiatives
16
Handles Resolve to Typed Data
Just one example - also looks like a digital
object
Handle Record
Handles can also have semantics but we frown on
it! Resolution is independent of semantics in
every instance
17
Allocation of Prefixes

1. - System Uses
2. - High Fan in/out Organizations
4. - Businesses and formal organizations
6. - Individuals and anything that cant fit above

18
Creating Resolving Type Information Dynamically

• Prefixes of the form 0.X are reserved for
  defining resolvable system information such as
  types and naming authorities
• 0.type/lttypegt is a handle for the type in
  brackets
• 0.na/ltnagt is a handle for a particular na
• Non-system types can also be created by
  individual users

19
Global Handle Resolution
HANDLE ADMINISTRATION
HS1
HS2
HPS3
HS4
Multiple Handle Servers
Handle Servers
HANDLE RESOLUTION
(Handles are uniformly spread by hashing)
20
Global Local Handle Resolution
HANDLE ADMINISTRATION
HS1
HS2
HPS3
HS4
Global
Handle Servers
HANDLE RESOLUTION
Local
HANDLE RESOLUTION
21
How do handles resolve...
1. Where is 1895.22/1011?
GHR
LHS C
Map of LHS B
LHS A
LHS ..n
LHS D
Handle Client
LHS B
2. Give me all data for 1895.22/1011
Handle Data
Handle System

• Two steps to resolve a handle - -
• Client queries GHR Which Handle Service has
  1895.22/1011?
• GHR responds with a map showing the client
  which servers within the responsible LHS it can
  query for that handle .

22
Administration of Handle Records
univ/thesis.txt
1217/4913527
univ/4913527
1217/thesis.txt
(the handles shown above identify digital objects)
univ
1217
1217.34
univ.csl
univ.csl.17
1217.34.1
univ.csl.17.2
23
The Global Handle Registry
Global Handle Registry
MIT Handle Service
Twin Bays Handle Service
DOI Handle Service
Korean Ctrl Lib Handle Service
LOC Handle Service
CMU Handle Service
Liqid Krystal Handle Service
Natl Lib Australia Handle Service
DTIC Handle Service

• The GHR is a unique handle service used to store
  the identity and location of all local handle
  services (LHS), and tells a handle client which
  service to query to resolve a handle.
• All handle clients (for resolution or
  administration) know how to contact and query the
  GHR.

24
Groups of Handle Servers
P
S
S
Group A
Group B
S
S
Group C
Group D
25
Handle Clients

• Administration

Use the Java Handle Client Tool provided in the
distribution for creating or updating handles
one-at-a-time or via a batch.
or
Develop your own administration client.
26
Handle Clients

• Resolution

Download web browser plug-in which enables
browsers to recognize the handle protocol.
or
Append a handle to proxy server e.g
http//hdl.handle.net/lthandlegt) which understands
both HTTP and HDL protocols.
or
Develop your own resolution client.
27
Setting up a Local Handle Service...

• Download the software from http//www.handle.net.
• Follow the instructions in the installation
  script.
• Send your site bundle, containing the IP
  address of your server and your administrator
  information, to the Global Handle Registry (GHR)
  administrator.

28
Organization of the International DOI Foundation
Members are Mostly Book Journal Publishers ?
Membership Dues
IDF

• Policies Procedures
• Licensing the DOI TM
• Qualifying RAs
• Marketing the DOI brand

Registration Agencies (RA)
CDD
IDF is a non-profit organization with offices
in Washington, DC (AAP) Geneva, Switzerland
(IPA)
4 per DOI on deposit 1X min 20K/yr 1 per
DOI in CDD on 12/31 annual ½ per DOI in CDD
after 50K per RA
29
Business Potential

• Enabling new forms of Creativity
• New forms of expression
• Representing value as Digital Objects
• Selling infrastructure technology services
• Enabling Third Party value-added capabilities
• Helping organizations manage their own
  information better offer new types of services
• Stimulating access to surface information and
  embedded information with appropriate access
  controls and conditions of use

30
Evolution of Policy for Global

• Original Policy
• Best efforts service run in-house
• Cost paid by the Government
• Available to the research community for free
• Current Policy (still in flux)
• Best efforts service run 7x24 with backup
• Free to the research community commercial users
  pay after a period of experimentation
• Handle System Advisory Committee oversees costs
  and evolution.

31
Cost of Global Services

• IPv4 ? several million addresses about 50M TLDs
  (excluding CCs)
• At say 20 per year per TLD, the cost of global
  registration and resolution services is about 1B
  per year this is inefficient, very profitable
  or both
• The handle system is almost as large as DNS
  (there are over 10M DOIs alone) and costs about
  250K per year at present.
• The DNS can be run within the handle system, if
  desired but the handle system can support IPv4
  and IPv6 without DNS

32
Applications of the Technology

• Identity Management (DHS)
• PKI Infrastructure
• Personal Locator Information
• Efficient Communications
• Steganography
• Managing Digital Cash
• Managing Business Transactions (e.g. email)
• Learning of more up to date Publications
• Cataloguing and Indexing