architecture - Domain Model - Identifier Relationships vs. Hierarchal Objects -

While extracting a fictional domain model, I myself am thinking that what would be a better way to relate domains to objects The domain object has a pointer (or child identifier), or it would be a better way to use a child object to create a composite inside the original object.

I can see the pros and cons of each approach, most of the business closure, which bubbles down, compared to the complexity of the shape, I tend to bend towards an identifier relationship approach because I have some kind of I do not anticipate the lazy load requirement.

Although not directly related, domain objects are simple POCOs (.NET-POOSO) marked them as serials, because there is a good chance that they will eventually cross between the application domain. .

Any ideas will be appreciated

Edit: A couple of more ideas can lean towards the identity-only approach. First of all, Policy. It is entirely possible that parents and children have different TTLs, as defined by their policies. The second would be that the referencing object can limit the reuse, in which case the same child can be organized by many parents in case of reusable data. Both of these are also related to the overall size of the ordered items.

I use both on my POCOs Generate with a basic example on the list and hair class.