0

I and my team are all beginners with NoSQL, were still wearing the Entity Framework with SQL Server 2008 on a project but with the passage of time the project was getting bigger and more complex than the EF does not resolve more for us, we decided to adopt the MongoDB, but we still have many doubts due to great paradigm shift, I'll post them here to see what you guys think and your opinions.

I have the entities "Person Fisica", "Patient" and "professional" and the patient and the Professional are Person, but in a few moments the patient and the professional will be the same person ex (a professional health unit which is also patient) in SQL Server we had a patient who had a reference to the Physical Person and professional who also had a reference to the PersonWhen Patient and Professional were the same person, the two had references to the same Person, now at mongo appeared doubts, some team members here want to do the same thing pretty much, Patient and professional organizations have the Id of the Person. Now I wanted to make the patient and the professional have the full object Person, but oh how would the integrity of this? Because technically the Physical Person of the Patient would be different from the Physical Person of professional ... This and other questions are breaking our heads here, in several entities that are shared do not know if we put the entity within the object that has it or the object only takes the Id of the entity, in the same way as in relational DB. Another example: the Health Unit and the types of UnidadeDeSaude, a type Of Health Unit has several Health Units and a Health Unit has a type, the correct approach would be to place the Unit Type object within the Health Unit or just reference it by Id?

Googled a few articles, but we're still in doubt in these cases http://highlyscalable.wordpress.com/2012/03/01/nosql-data-modeling-techniques/ http://blog.fiesta.cc/post/11319522700/walkthrough-mongodb-data-modeling

Improve this question

1 Answer 1

Reset to default
1

Without being able to see exactly what you have, so speaking generally, in MongoDB you wouldn't JOIN tables in the same way you would with a RDBMS. Typically, if you have a Person entity you store the whole Person as a Person. This is a nice mapping from your code classes.

In the case where you have references to other entities, say where a single Person is shared between Patient and Professional you would do this with a foreign key reference in a RDBMS. You can do this with Mongo but Mongo won't do the JOIN for you. That would have be done by the caller. The recommended approach is to put a copy of the Person entity in both Patient AND Professional. The impact of this means that if you update the Person entity you now have to update the data in two places, but that's not necessarily as bad as it sounds. It's usually "quick" to update and you can update both 'atomically' so in practice there's little difference between that and updating a single entity except that you don't have to do the JOIN so your reads are simpler and usually faster.

The most powerful tool you have for fetching data is the Collection's (table's) index over your documents (entities) and any way you can leverage that will be the fastest way to return data. So counter intuitively, if you need to filter and process parts of a document more often than you need the whole, you are better to break it up into entities that share an indexed key. That would mean storing Person, Patient and Professional in the same collection and using two keys. One key is shared by the Person and it's derived class (Patient) and the other is a type discriminator that selects one part or the other. In other words, use the index to find whole entities, or collections of whole entities.

Aside from that, if once you have used the index to locate an entity, Person, Patient or Professional, read the whole entity and have it contain everything you need to fulfill the request without a JOIN. So whether you request the Patient or the Person (both refer to the same Person) you get the same Person data whichever object you read.

In short, you'll be replicating data in Mongo just about anywhere you'd have used a Join in SQL.

Are you able to draw what your class hierarchy looks like?

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

OK, so Professional and Patient don't derive from PhysicalPerson but contain one. Unlike what you've been doing with an ORM, in Mongo, you would just serialize the Patient as it is, including the Person object embedded directly, assuming you want to read the whole Patient in one operation. Remember to do a JOIN that would be two operations. However, if you want to operate on Person objects separately then they would be better off as separate top-level Entities in a collection. In fact you can do both, depending on what suits your queries best.
Thanks @cirrus, that clarified a lot of things for me and my team!

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.