DB Design for dynamic data objects

Hi

I have an interesting design issue.

I want to create a database design were a custom type objects can be created, linked and retrieved.

I am planning to design a framework which will provide user a capability to define his own object structure and store it. He should be able to retrieve the values based on the fields defined in his objects.

I am planning to have XML to pass the information between client and the server, but I am struggling with coming up with an acceptable database design for this.

I want to design DB considering storing capacity of upto 40,000 records and custom objects created upto 200.

Any ideas or guldens will be highly appreciated.

Thanks

If your database consists only of a collection of user-defined objects, then the only person who can access it is the user who created it, because the only framework is the implied one between values in fields which are equated to one another. If you are looking for a framework which connects collections of objects, possibly from different users and allows searches to be conducted then you are looking at a framework which sits outside the actual user-defined objects but which contains all of the relationships and various details of the object structure which allows queries to be formulated. This is called metadata ("data about data" in the trade.
The basic table will be a list of objects containing only identification data and type data. It will also contain pointers to one or more records containing the actual object data. You will then need a one or more tables to describe the self-referential relationships between objects of given types. For any given object type, you will need one or more tables which tell you how to get into the object and look for matching data. The nature of the this data will depend upon the constitution of the object. For instance, table-based objects will simply have field names, but complex objects such as BLOBS (Binary large objects) or memos (free-form text with many thousands of characters) may need something different.
In a post to another thread in this forum ("Handling Data For Different Customer Needs", I provided a schema for business purposes, where the user could define their own tables for things like ordering, staff management and so forth, but they were all connected by a module containing tables describing the real-world entities which were participants in the transactions and relationships. You might find it useful to extend your train of thought.
The problem with any project like this is that any framework has to have some idea of what sort of objects are going to be defined, so that the user can enter the metadata which will allow his/her objects to interface with other objects.

After posting my earlier thoughts on this topic, I began to wonder exactly what is meant by "dynamic data objects". If it simply means tabular data wherein the content changes frequently, then my post applies. However, it occurred to me that one could imagine an object which is changing constantly. For instance, a log of a supernova, or an artwork to which multiple artists are contributing. One would then need to extend the metadata model by inserting a status variable into the entity table and an event history of some sort to chart the changes.