Web Storage: Permanence of Information

1
Web Storage Permanence of Information

• By Laura Milodin
• IST 497

2
Overview

• Introduction
• Problems
• Solutions
• Some alternatives
• Conclusion

3
Introduction

• Through publication we preserve and transmit our
  knowledge and culture.
• Electronic media offers clear advantage to
  transmit knowledge.
• However preserving our knowledge offers some new
  challenges.

4
The Problem

• At some time in the future, we may want some
  information that we have today.
• We want that information to be efficiently
  available to us in the future.

5
Web permanence

• Narrow sense -- we have control of a particular
  piece of web content which we want to remain
  accessible to Web users.
• Broader sense -- we want to save everything of
  value on the Web in order to preserve our
  culture.
• The information presented today is looking at the
  problem in the narrow sense of Web permanence.

6
Actions we might have to take

• 1. We have to protect from unexpected disaster.
• 2. We have to protect from known slower acting
  deterioration.
• 3. We have to keep the content accessible to
  users.
• 4. We have to maintain not distort the importance
  of the content stored.

7
Importance of the content

• Whatever value society places on something should
  not be affected by how we store it.
• We might however store something differently
  because of its importance.
• The storage process itself should do nothing that
  would prevent recovery of important content in
  preference to less important.

8
Distortion of importance

• Hardware needed for access may become less
  available over time, thus favoring content that
  is read by newer equipment.
• Some content might be easier to access than other
  more important content.
• Accidental discovery of some content might be
  much more likely than for some more important
  content.

9
Solutions

• Intermemory is a noncommercial, decentralized
  concept using a scheme similar to barter.
• An alternative to commercial archives or archival
  services offered by large libraries.

10
What is Intermemory barter?

• There is no central organization, rather each
  subscriber donates a certain amount of storage
  space for a limited time and in return receives
  the right to archive a much smaller amount.
• Hard numbers are difficult to calculate due to
  many parameters.

11
Intermemory basics

• The Intermemory is a very large, distributed,
  self-organizing memory consisting of the combined
  memory of all subscribers that is addressed by a
  single addressing scheme.
• The addresses correspond to blocks of N words
  each of w bits each.

12
How it Works?

• Redundancy and dispersal provide the protection
  from unexpected disaster.
• The rebuilding of data from a discontinued
  subscriber by a new subscriber provides the
  updating of equipment for the whole system. If
  one processor fails, its data is reconstructed
  automatically by a new or other existing
  subscriber.

13
Redundancy and Dispersal

• A particular data block exists in its entirety
  only at a single processor.
• The address of this block is used to retrieve the
  data under normal circumstances.
• The portions of the data in this block are also
  dispersed among many other processors, this is
  used to rebuild the original data in case of
  failure.

14
Space-optimal dispersal

• The mathematics behind the dispersal algorithm
  involve polynomial evaluation and interpolation.
  It is based on the idea of associating every
  block of N words with a polynomial of degree N-1.
  The value assumed by this polynomial at N
  distinct points would uniquely identify it.

15
Retrieval

• We calculate more points than we need, say 2N
  points.
• We disperse these 2N points among many
  processors.
• If we lose the original word block, we only need
  to recover N out of possible 2N points to
  recover the polynomial and find the original
  block that corresponds to it.

16
Space requirement

• Looking at space requirements, at the first level
  of replication, each dispersal level takes twice
  the original block size. At the second level of
  replication takes four times the original block
  size.
• Total space requirement is 9 times the original.

17
Degree of dispersal

• The degree of maximum dispersal is a key variable
  here.
• Dispersal on the scale proposed in the former
  slide would not be needed for a model where
  processor failures were independent.
• This model assumes possibility of software, bugs,
  viruses, overt adversarial action.

18
Uniform Resource Locators

• Another alternative is the URL.
• The goal here is not to physically maintain
  hardware, or a readable copy of content on that
  hardware, but rather to maintain a link.
• We are assuming that a readable copy exists we
  want to be able to link to it even when its
  physical location or file structure changes.

19
Uniform Resource Name, URN

• The general solution to this problem is the
  development of Uniform Resource Names or URNs.
• A parallel situation exists with books in a
  library. We could attempt to describe a book as
  on the fourth floor, third aisle from west end,
  top shelf, second from end. Or we could give the
  book a name and maintain some way of resolving
  the name into its location.

20
Persistent URL (PURL)

• PURLs are one possible solution to the problem
  of developing URNs.
• PURLs look and function much like URLs, but
  instead of pointing directly to the location of
  an Internet resource, a PURL points to an
  intermediate resolution service.
• Ex PURL http//purl.oclc.org/NET/frankdill/index.
  html
• URLhttp//members.aol.com/aiaio/index.html

21
PURL Server

• The process of resolution now involves one extra
  step in which a PURL server associates the PURL
  with a unique URL which is returned to the
  client.
• The extra step is an HTTP redirect
• The key to PURL server is indirection, not
  redirection, naming something to separate
  location from identification.

22
PURL database

• PURLs must be maintained.
• A change to the PURL database is required if the
  owner of a file moves that file.
• If file is completely removed, of course, the
  PURL is as useless as the URL.

23
Some alternatives

• Meanwhile, there is a website that archives all
  the web pages that posted on the World Wide Web.
• The name of it is Internet Archive .
• The URL for it is http//www.archive.org .

24
Conclusion

• Introduction of web storage
• Problems
• Solutions
• Some alternatives for right now

25
References

• Towards an Archival Intermemory by Goldberg
  Yianilos http//citeseer.nj.nec.com/goldberg98towa
  rds.html
• Introduction to Persistent Uniform Resource
  Locators
• by Shafer, Weibel, Jul Fausey
  http//purl.oclc.org/docs/inet96.html
• Web Storage The Permanence of Information
• by F. Dill http//external.nj.nec.com/homepages/k
  rovetz/timetable.html

26
(No Transcript)
27
Any Questions?