Alex Rivera | Logout

How would I model data that is heirarchal and relational in a document-oriented database system like RavenDB?

Asked 2011-06-08T21:51:36.667
12

Document oriented databases (particularly RavenDB) are really intriguing me, and I'm wanting to play around with them a bit. However as someone who is very used to relational mapping, I was trying to think of how to model data correctly in a document database.

Say I have a CRM with the following entities in my C# application (leaving out unneeded properties):

public class Company
{
    public int Id { get; set; }
    public IList<Contact> Contacts { get; set; }
    public IList<Task> Tasks { get; set; }
}

public class Contact
{
    public int Id { get; set; }
    public Company Company { get; set; }
    public IList<Task> Tasks { get; set; }
}

public class Task
{
    public int Id { get; set; }
    public Company Company { get; set; }
    public Contact Contact { get; set; }
}

I was thinking of putting this all in a Company document, as contacts and tasks do not have a purpose out side of companies, and most of the time query for a task or contacts will also show information about the associated company.

The issue comes with Task entities. Say the business requires that a task is ALWAYS associated with a company but optionally also associated with a task.

In a relational model this is easy, as you just have a Tasks table and have the Company.Tasks relate to all tasks for the company, while Contact.Tasks only show the tasks for the specific Task.

For modeling this in a document database, I thought of the following three ideas:

  1. Model Tasks as a separate document. This seems kind of anti-document db as most of the time you look at a company or contact you will want to see the list of tasks, thus having to perform joins over documents a lot.

  2. Keep tasks that are not associated with a contact in the Company.Tasks list and put tasks assocaited with a contact in the list

data-modeling ravendb document-database document-based-database
Edit
Report

1 Answer

10

Use denormalized references:

http://ravendb.net/faq/denormalized-references

in essence you have a DenormalizedReference class:

public class DenormalizedReference<T> where T : INamedDocument
{
    public string Id { get; set; }
    public string Name { get; set; }

    public static implicit operator DenormalizedReference<T> (T doc)
    {
        return new DenormalizedReference<T>
        {
            Id = doc.Id,
            Name = doc.Name
        }
    }
}

your documents look like - i've implemented the INamedDocument interface - this can be whatever you need it to be though:

public class Company : INamedDocument
{
    public string Name{get;set;}
    public int Id { get; set; }
    public IList<DenormalizedReference<Contact>> Contacts { get; set; }
    public IList<DenormalizedReference<Task>> Tasks { get; set; }
}

public class Contact : INamedDocument
{
    public string Name{get;set;}
    public int Id { get; set; }
    public DenormalizedReference<Company> Company { get; set; }
    public IList<DenormalizedReference<Task>> Tasks { get; set; }
}

public class Task : INamedDocument
{
    public string Name{get;set;}
    public int Id { get; set; }
    public DenormalizedReference<Company> Company { get; set; }
    public DenormalizedReference<Contact> Contact { get; set; }
}

Now saving a Task works exactly as it did before:

var task = new Task{
    Company = myCompany,
    Contact = myContact
};

However pulling all this back will mean you're only going to get the denormalized reference for the child objects. To hydrate these I use an index:

public class Tasks_Hydrated : AbstractIndexCreationTask<Task>
{
    public Tasks_Hydrated()
    {
        Map = docs => from doc i
answered 2011-06-09T08:22:27.383

Your Answer