Distributed database: Difference between revisions
No edit summary |
No edit summary |
||
| Line 7: | Line 7: | ||
http://www.hl7.org/library/committees/structure/minutes/Minutes041999.doc | http://www.hl7.org/library/committees/structure/minutes/Minutes041999.doc | ||
*[[object-oriented database]]s like [[ORB]] to make the data more consistently available via one programmatic interface; | *[[object-oriented database]]s like [[ORB]] to make the data more consistently available via one programmatic interface; | ||
*[[relational database]]s that further structure the data but restrict data types to a fixed set; Typically these are quite complex to truly distribute | *[[relational database]]s that further structure the data but restrict data types to a fixed set; Typically these are quite complex to truly distribute, but see [[relational state transfer]] for how this might be done with help from: | ||
*[[peer to peer]] networks that are more strictly hierarchical and reliant on [[namespace mechanics]] to match finder and publisher, e.g. [[Overnet]], some [[REST]] equivalents now in early testing | *[[peer to peer]] networks that are more strictly hierarchical and reliant on [[namespace mechanics]] to match finder and publisher, e.g. [[Overnet]], some [[REST]] equivalents now in early testing | ||
Any of these might be required in [[Distributed Consumerium]] | Any of these might be required in [[Distributed Consumerium]] | ||
Revision as of 18:47, 7 April 2004
There are several forms of distributed database:
- the normal web in which URIs change often and are inconsistent, e.g. across wiki code releases, which is really inexcusable given mod_rewrite
- standard wiki URIs making uniformly licensed and formatted articles available, e.g. public domain interwiki or GFDL corpus wikitext standard collections, at URIs which are easy to remember and persistent
- semi-structured wikitext with interwiki link standards and interwiki identity standards that make it possible to specify "what" and "who" in a way that works across an entire corpus - this would create a true GFDL Corpus
- standardized XML DTD and XML-Schemas, e.g. as proposed for health care
http://www.hl7.org/library/committees/structure/minutes/Minutes041999.doc
- object-oriented databases like ORB to make the data more consistently available via one programmatic interface;
- relational databases that further structure the data but restrict data types to a fixed set; Typically these are quite complex to truly distribute, but see relational state transfer for how this might be done with help from:
- peer to peer networks that are more strictly hierarchical and reliant on namespace mechanics to match finder and publisher, e.g. Overnet, some REST equivalents now in early testing
Any of these might be required in Distributed Consumerium