domain-driven-designbounded-contexts

Should a complex aggregate in one bounded context be aggregate in another bounded context?

There are two bounded contexts Context1 and Statistics.

Context1 contains SomeEntity with complex structure SomeEntity 1---* EntitiesLevel1 1---* EntitiesLevel2

Simple versions (without some fields) of SomeEntity, EntitiesLevel1 and EntitiesLevel2 are needed in Statistics context because they are used in buidling of some graphs in this context.

When changes happen in entities EntitiesLevel1 or EntitiesLevel2 (through SomeEntity) in Context1, events are sent to Statistics context.

Should I update EntitiesLevel1 and EntitiesLevel2 in Statistics context throught SomeEntity (in another words, implement in Statistics context almost the same Aggregate as in Context1 but without some fields) or SomeEntity, EntitiesLevel1 and EntitiesLevel2 should be treated as just Value Objects which can be updated separetly without using Aggregate pattern?

I'm not sure because from one side SomeEntity, EntitiesLevel1 and EntitiesLevel2 still keep their hierarchy in Statistics context and because of that they can be updated only in way that Aggregate pattern defines, but from another side these Entity.* in Statistics context are used for graphs only and Statistics context cannot change them (only handlers of events from Context1)

Solution

  • Aggregates protect transactional boundaries, ensure invariants are never broken, and represent domain concepts in code. None of which seem to be necessary or meaningful in your statistics context.

    So whether you duplicate the structure in the Statistics context or replace it with something else, you should mentally disassociate (and actively avoid) connecting the aggregates in Context1 with the Statistics context. The terms "Aggregates", "Value Objects", and "Events" do not make sense in your Statistics context.

    The read-only data structures you plan to construct in the Statistics context are just that - read-only models that can have any format but are built with only one goal in mind - to aid querying in specific ways later. You are free to use any format or structure that makes it easier to satisfy the requirements in the Statistics context.