Internet Public Library

     The Online Services Development Association (OSDA)

                      a proposal for a

       PUBLIC access RESOURCE DATABASE NETWORK (PRDN)
       ==============================================

                        (c) MARCH 1983


WHAT IS OSDA:

This  is simply a working group title for a number of  sysops
and other individuals who have been active in the development
online  services. OSDA is a non profit organisation and has a
good   working  relationship  with  a   number   of   helpful
organisations.


INTRODUCTION:

The  general  concept of a distributed database is  not  new.
However   this   proposed  application  of  the  concept   is
different,  innovative  and more important,  significant  for
what it may achieve.

Simply,  the proposal is to develop a low cost, Public access
Resource Database Newtwork (PRDN).


Sounds expensive...?

We  hasten  to  add that due to the different  and innovative
nature of this proposal, it is not so.

Perhaps  a  suprise to some, a great deal of the initial  and
innovative ground work has been quietly going on for a number
of years.

Much  of the credit being due to many dedicated hard  working
volunteer individuals and organisations including enlightened
elements in the computer industry.


WHY A PRDN:

There  are a great many important and beneficial reasons  why
the  proposed PRDN should be developed, to briefly touch on a
few:

There  is a very real danger that 'Information Poverty'  (IP)
will  continue  to develop along side our apparent  ecconomic
ills   leading  to  more  immense  and  difficult  to   solve
developmental problems in the future.

A  developed PRDN would make a very significant  contribution
to  development in all areas. This being made possible by the
availability of useful information and resources.

There  are still tremendous perceptual problems  amongst  the
population at large about the computer's role in society. Few
fully appreciate the computer as a useful tool.

A  suitably presented PRDN could do a great deal to promote a
wider  public understanding and acceptance of the benefits of
computers as a useful tool.

The  greatest wealth of useful information is held by society
as  a whole. A PRDN can tap this COMMON RESOURCE and begin to
make it AVAILABLE and AFFORDABLE to all.

Parallel  benefits  to  the development of a  PRDN  would  be
ecconomic  development for makers and providers of  requisite
equipment,   software,  services.  The  tradititional   costs
associated  with obtaining Information and Resources could be
reduced dramatically.

By providing for the user to contribute useful information to
the PRDN, purposeful, creative and satisfying employment will
be  possible  for  many. This would also do much to  build  a
sense  of belonging and community into fragmented elements of
society.


WHY NOT A CENTRALISED SYSTEM:

Generally,  traditional systems are centralised, which  often
means  relatively  high overheads, high investment  and  slow
adaption to change, thus they can be obsolete very quickly.

The  methods  of  networking (information  distribution)  use
currently  expensive high speed data links to  provide  quick
(real time)  access  to  the  distant  centralised  store  of
information.

Such systems command relatively high service charges which in
themselves limit wider access.

Information  is  often  kept  secret  and  protected  against
open publication so its value stays high to ensure sufficient
return.

This  may  be considered good business practice, but  it  can
also  slow  the  wider and more  general  development  of  an
efficient society and feed Information Poverty.

To  put  such  systems  in  perspective,  while  they  remain
relatively   expensive  to  use,  they  are  best  suited  to
specialist  information services for which there is little or
no need by the majority.


DEVELOPING A SUITABLE PRDN:

There  is already an embryonic backbone of such a system with
a number of willing system owners and operators.

This  backbone  is made up of a number  of  Nodes  (computers
with  substantial  data storage  facilities).  These  already
network  with each other to exchange information,  electronic
mail  (email) and data, mainly via the telephone system on  a
need to connect basis (ie: local node information changes  or
new email requiring delivery automatically triggers the  node
to make appropriate connections on a priority basis).

Collectively  these individual computers  contain  a  growing
wealth of useful information.

A  simple  to  use  automatic  RESOURCE  DATABASE  system  is
required to enable Nodes to provide access to the information
on the collective system.

There  are  a  number  of ways of  achieving  this,  what  is
required is general agreement on the best method.


RESOURCE DATABASE OUTLINE

One method is as follows:

The  Resource Database would provide a simple menu system  to
provide subject, title, description and keyword search of the
collective information.

Immediate  access would be given to information held  on  the
local node,  while  other  information  would  be   requested
from  the node holding the information. Subject  to  priority
this would be transfered immediately or later.

Regularly requested information could be held locally to save
on  networking overheads. Local Node  software  automatically
deciding what to hold localy.

All  nodes would use the same Resource Database  files,  with
changes  and  additions automatically updated  by  a  central
processing node with at least one full backup node.

Individual  nodes  would  automatically  forward  details  of
changes to their content for inclusion.

It  would  not  be  neccessary  to  redistribute  the  entire
database  as Node based software would  automatically  update
the main database files from small Content Change files.

As  a PRDN develops and traffic increases faster  data  links
may become economic, providing wider real time access.


Some  suitable  form of reimbursment to cover  costs  may  be
considered neccesary. It is important that whatever method is
introduced, it does not prevent those most in need from using
such a system.

Some  Sponsorship  Funding may be found by  encouraging  some
forms of advertising. Within limits the General Public  might
be freely permited to Advertise Items for Sale while business
advertising  could  provide  sponsorship.  An  ADvert   INDEX
(Adindex)  could  enable  fast  location  of   items   needed
and improved use of resources.

Teachers  frequently  reinvent the  wheel  when  they  create
teaching material their colleagues have done countless  times
before.  Such resources could be put in an Education  Library
with open access and contributions from all.

It  may  also  be appropriate to  include  the  facility  for
Authors of material to charge some suitable small amount each
time  their  information is transfered to an End  User.  This
could greatly encourage the provision of useful  information,
electronic publications and software.





            SOLUTIONS TO SOME TECHNICAL PROBLEMS
            ====================================




VOLUME OF DATA:

With  low cost Gigabyte storage for computers just round  the
corner we need to plan for a large amount of potential growth.

While  a PRDN is embryonic the index files will not be large,
perhaps  taking  up megabytes. However when  considering  the
techniques  we  innovate or adopt, we should keep an  eye  to
the future growth to avoid later problems.

Longer  term,  we  should  perhaps expect a  minimum  of 1000
gigabytes of storage within a national PRDN (It is  difficult
to estimate this).

Allowing  for  localised duplication of files held  on  other
Nodes  we  may  expect  the main INDEX  files  to  cope  with
5,000,000  file titles and descriptions in the not to distant
future.


INDEX DATA STRUCTURE:

We need something fairly simple and compact in use of storage
space and network transfer time.

A Subject number system would save considerable space,  three
bytes being required for 16,777,216 subject areas. The  local
Node would use a look up table for conversion.


1. SUBJECT (3 byte number with look up table)
2. TITLE (word number token string)
3. KEYWORDS (word number token string with look up table)
4. FILE DESCRIPTION (compex compresed string data)
5. INFORMATION STATUS (one byte with look up table)
6. PRIMARY LOCATION (holding node address 4 byte number)
7. SYSTEM FILE NUMBER (Read/write name 11 bytes)
8. FILE SIZE (2 byte number nearest K)


The  above  index format uses the least space  but  may  need
modification  if data processing time is likely to  become  a
problem.


COMPRESSED TEXT STRINGS:

The  COMPEX protocol permits fast real time  compression  and
decompression of string text.

This  works  on  the basis that about 75% of text  uses  only
thirteen or so characters from the alphabet. It uses a nibble
code  algorithum  to code and decode text  accordingly.  This
could  provide  a reduction in size of between 50%  and  30%,
subject to the use of characters.


COLLECTIVE SYSTEM FILE NUMBER:

To maintain downwards compatability we will be stuck with the
eight  character file name and three character type part used
by many systems.

A suitable number scheme is to use characters "0" through "9"
and "a" through "z" to provide a 36 base number system.  This
providing for around

                   130,000,000,000,000,000

unique file names, sufficient for 26,000,000 file names  each
for every member of the worlds population.


COMPLETE SYSTEM ACCOUNTING:

Complete  system  accounting  with  the  ability   to   check
costs, level  of use for information provided,  files,  Users
and traffic would be essential.