More ‘Blog as a Database’ Thoughts
Listening to: Push Upstairs - Underworld
While I was ironing my shirt this morning (hey! looking this
good doesn’t happen by itself!) I pondered the concept of using current blog software
as a database engine. I guess one of the characteristics of traditional databases
is the ability to describe relationships between entities. I figure that if you’re
happy with a combination of unstructured data entities and loose relationships then
it’s currently feasible to use a blog as a database engine.
Let’s take something like the venerable old ‘pubs’ database that ships with SQL Server:
What you would do is set up six blogs:
- Stores
- Sales
- Titles
- Publishers
- Employees
- Authors
Each would have a list of postings, one posting per record. To link each of the entities
together you would simply link the appropriate records using the permalink for each
posting. The permalink is sufficient because it is something unique enough (GUID)
to prevent clashes on a world wide level so it should be pretty good for a small database.
OK, an example!
If I were an author who had written 3 books there would be 4 postings (3 in the Titles
blog and 1 in the Authors blog). Each of the book postings would have an href
link to my author posting (to speed up queries my author posting might have links
to the book postings… but if I had Trackbacks turned on these linkages would emerge
over time anyway). Super simple.
Note that I didn’t add a TitleAuthor blog? It seems to be redundant in this environment.
Changing, adding and deleting records would be done using a custom interface (not
much more than a RSS posting tool that understood about relationships) that used something
like the Blogger API to write to the appropriate blog.
The benefits as I see them:
- It’s based on the nicely redundant web
- You can access all ‘tables’ using their URL (eg. some-domain.com/pubs/Titles)
- Tables will render themselves in a human readable manner
- It’s simple
- Non-technical users should ‘get it’
-
It’s sloppy… but then all killer standards are … it makes it easy for real
people to use them because the systems are more tolerant
Downside
-
Not too sure how it would scale as the number of entries grew… there are blogs will
1000s of posts, but I don’t know of any with millions -
Complex queries would need to occur in code *retch* (but that’s until there was a
sufficiently useful query engine built that allowed the separation of data querying
and data manipulation)
UPDATE: Got a good question from Doogie Howser about referential integrity. That is
one of the other downsides… that would need to be enforced at a code level.
