Using LDAP as an alternative for SQL trees

[SomeDude]

Lo group,

During my quest for the "perfect" way for storing tree-structures it
quickly became clear that all of the available SQL-solutions have their
own peculiar weaknesses (scalability, slow querying, dangerous inserts,
etc).

This makes me wonder if storing a tree-structure in LDAP would be a
realistic alternative? Scaling probably wouldn't be a problem but how
about speed and stuff?

To make things clearer I am thinking about retrieving all
path-info for my webbased application (up to the leafs which can be
comments, posts, links or whatever) from the LDAP server and then loading
the related data from a database.


Any insights/experiences would be highly appreciated,


SomeDude.

[Chung Leong]

Re: Using LDAP as an alternative for SQL trees

I don't know what you mean by perfect. If I have a data structure I
want to save, I'd serialize it and store it somewhere, either in a file
or in text column in a database, so that I could recreate it in memory
later.

[troelskn]

Re: Using LDAP as an alternative for SQL trees

Have you considered using an xml-database ?

http://www.rpbourret.com/xml/ProdsNative.htm

[SomeDude]

Re: Using LDAP as an alternative for SQL trees

On Thu, 08 Sep 2005 06:57:36 -0700, Chung Leong wrote:
[color=blue]
> I don't know what you mean by perfect. If I have a data structure I
> want to save, I'd serialize it and store it somewhere, either in a file
> or in text column in a database, so that I could recreate it in memory
> later.[/color]

With perfect I mean an alternative way for storing large trees using a
method that will outdo the following common methods:
- adjacency method (easy but query intensive)
- Materialized path
- Joe Celko Nested Sets (or MPTT which has slow updates)
- Tropashko's Nested Intervals with Continued Fractions (that does not
size well to both fat and high trees)