I am creating the high level design for a new service. The complexity of the service warrants using DDD (I think). So I did the conventional thing and created domain services, aggregates, repositories, etc. My repositories encapsulate the data source. So a query can look for an object in the cache, failing that look in the db, failing that make a REST call to an external service to fetch the required information. This is fairly standard. Now the argument put forward by my colleagues is that abstracting the data source this way is dangerous because the developer using the repository will not be aware of the time required to execute the api and consequently not be able to calculate the execution time for any apis he writes above it. May be he would want to set up his component's behaviour differently if he knew that his call would result in a REST call. They are suggesting I move the REST call outside of the repository and maybe even the caching strategy along with it. I can see their point but the whole idea behind the repository pattern is precisely to hide this kind of information and not have each component deal with caching strategies and data access. My question is, is there a pattern or model which addresses this concern?
They are suggesting I move the REST call outside of the repository
Then you won't have a repository. The repository means we don't know persistence details, not that we don't know there is persistence. Every time we're using a repository, regardless of its implementation (from a in memory list to a REST call) we expect 'slowness' because it's common knowledge that persistence usually is the bottleneck.
Someone who will use a certain repository implementation (like REST based) will know it will deal with latency and transient errors. A service having just a IRepository dependency still knows it deals with persistence.
About caching strategies, you can have some service level (more generic) caching and repository level (persistence specific) caching. These probably should be implementation details.
Now the argument put forward by my colleagues is that abstracting the data source this way is dangerous because the developer using the repository will not be aware of the time required to execute the api and consequently not be able to calculate the execution time for any apis he writes above it. May be he would want to set up his component's behaviour differently if he knew that his call would result in a REST call.
This is wasting time trying to complicate your life. The whole point of an abstraction is to hide the dirty details. What they suggest is basically: let's make the user aware of some implementation detail, so that the user can couple its code to that.
The point is, a developer should be aware of the api they're using. If a component is using an external service (db, web service), this should be known. Once you know there's data to be fetched, you know you'll have to wait for it.
If you go the DDD route then you have bounded contexts (BC). Making a model dependent on another BC is a very bad idea . Each BC should publish domain events and each interested BC should subscribe and keep their very own model based on those events. This means the queries will be 'local' but you'll still be hitting a db.
Repository pattern aim to reduce the coupling with persistence layer. In my opinion I wouldn't risk to make a repository so full of responsibility.
You could use an Anti Corruption Layer against changes in external service and a Proxy to hide the caching related issues.
Then in the application layer I will code the fallback strategy.
I think it all depends where you think the fetching/fallback strategy belongs, in the Service layer or in the Infrastructure layer (latter sounds more legit to me).
It could also be a mix of the two -- the Service is passed an ordered series of Repositories to use one after the other in case of failure. Construction of the series of Repos could be placed in the Infrastructure layer or somewhere else. Fallback logic in one place, fallback configuration in another.
As a side note, asynchrony seems like a good way to signal the users that something is potentially slow and would be blocking if you waited for it. Better than hiding everything behind a vanilla, inconspicuous Repository name and better than adding some big threatening "this could be slow" prefix to your type, IMO.
Previously I asked this question and on a answer I got this comment:
This works, however injecting the container to a part, as far as I know, is not a "normal" use-case of MEF.
In my web app I have a few repositories that, of course, retrieve entities from the DB. To make them as loosely coupled as possible I'm making them return interfaces (eg IUser, IBill, IBlaBlaBla...) and the repository project only references the library project (that contains the interfaces). I use MEF composition capabilities to tie it all up...
Since the repository must have a concrete object to fill with info it got from the DB and the only the Container is aware of which concrete class maps to a specific interface I think that the repository MUST have reference to the container so it can call the "Resolve()", get a new instance and do his job, but that apparently is a mistake.
Can anyone tell me why and what approach would be better?
PS: I don't know if it's relevant but I'm using DDD...
I think the flaw in your design that lead to this problem is the use of interfaces to hide entities behind. Since entities are your core concept in your domain, there should be no use in hiding them behind an abstraction. Let's put it differently: do you ever have a different implementation of IUser?
In other words, ditch the IUser, IBill, etc. interface and let your repositories and business commands depend directly on your aggregate entities.
(Note: My question has very similar concerns as the person who asked this question three months ago, but it was never answered.)
I recently started working with MVC3 + Entity Framework and I keep reading that the best practice is to use the repository pattern to centralize access to the DAL. This is also accompanied with explanations that you want to keep the DAL separate from the domain and especially the view layer. But in the examples I've seen the repository is (or appears to be) simply returning DAL entities, i.e. in my case the repository would return EF entities.
So my question is, what good is the repository if it only returns DAL entities? Doesn't this add a layer of complexity that doesn't eliminate the problem of passing DAL entities around between layers? If the repository pattern creates a "single point of entry into the DAL", how is that different from the context object? If the repository provides a mechanism to retrieve and persist DAL objects, how is that different from the context object?
Also, I read in at least one place that the Unit of Work pattern centralizes repository access in order to manage the data context object(s), but I don't grok why this is important either.
I'm 98.8% sure I'm missing something here, but from my readings I didn't see it. Of course I may just not be reading the right sources... :\
I think the term "repository" is commonly thought of in the way the "repository pattern" is described by the book Patterns of Enterprise Application Architecture by Martin Fowler.
A Repository mediates between the domain and data mapping layers,
acting like an in-memory domain object collection. Client objects
construct query specifications declaratively and submit them to
Repository for satisfaction. Objects can be added to and removed from
the Repository, as they can from a simple collection of objects, and
the mapping code encapsulated by the Repository will carry out the
appropriate operations behind the scenes.
On the surface, Entity Framework accomplishes all of this, and can be used as a simple form of a repository. However, there can be more to a repository than simply a data layer abstraction.
According to the book Domain Driven Design by Eric Evans, a repository has these advantages:
They present clients with a simple model for obtaining persistence objects and managing their life cycle
They decouple application and domain design from persistence technology, multiple database strategies, or even multiple data sources
They communicate design decisions about object access
They allow easy substitution of a dummy implementation, for unit testing (typically using an in-memory collection).
The first point roughly equates to the paragraph above, and it's easy to see that Entity Framework itself easily accomplishes it.
Some would argue that EF accomplishes the second point as well. But commonly EF is used simply to turn each database table into an EF entity, and pass it through to UI. It may be abstracting the mechanism of data access, but it's hardly abstracting away the relational data structure behind the scenes.
In simpler applications that mostly data oriented, this might not seem to be an important point. But as the applications' domain rules / business logic become more complex, you may want to be more object oriented. It's not uncommon that the relational structure of the data contains idiosyncrasies that aren't important to the business domain, but are side-effects of the data storage. In such cases, it's not enough to abstract the persistence mechanism but also the nature of the data structure itself. EF alone generally won't help you do that, but a repository layer will.
As for the third advantage, EF will do nothing (from a DDD perspective) to help. Typically DDD uses the repository not just to abstract the mechanism of data persistence, but also to provide constraints around how certain data can be accessed:
We also need no query access for persistent objects that are more
convenient to find by traversal. For example, the address of a person
could be requested from the Person object. And most important, any
object internal to an AGGREGATE is prohibited from access except by
traversal from the root.
In other words, you would not have an 'AddressRepository' just because you have an Address table in your database. If your design chooses to manage how the Address objects are accessed in this way, the PersonRepository is where you would define and enforce the design choice.
Also, a DDD repository would typically be where certain business concepts relating to sets of domain data are encapsulated. An OrderRepository may have a method called OutstandingOrdersForAccount which returns a specific subset of Orders. Or a Customer repository may contain a PreferredCustomerByPostalCode method.
Entity Framework's DataContext classes don't lend themselves well to such functionality without the added repository abstraction layer. They do work well for what DDD calls Specifications, which can be simple boolean expressions sent in to a simple method that will evaluate the data against the expression and return a match.
As for the fourth advantage, while I'm sure there are certain strategies that might let one substitute for the datacontext, wrapping it in a repository makes it dead simple.
Regarding 'Unit of Work', here's what the DDD book has to say:
Leave transaction control to the client. Although the REPOSITORY will insert into and delete from the database, it will ordinarily not
commit anything. It is tempting to commit after saving, for example,
but the client presumably has the context to correctly initiate and
commit units of work. Transaction management will be simpler if the
REPOSITORY keeps its hands off.
Entity Framework's DbContext basically resembles a Repository (and a Unit of Work as well). You don't necessarily have to abstract it away in simple scenarios.
The main advantage of the repository is that your domain can be ignorant and independent of the persistence mechanism. In a layer based architecture, the dependencies point from the UI layer down through the domain (or usually called business logic layer) to the data access layer. This means the UI depends on the BLL, which itself depends on the DAL.
In a more modern architecture (as propagated by domain-driven design and other object-oriented approaches) the domain should have no outward-pointing dependencies. This means the UI, the persistence mechanism and everything else should depend on the domain, and not the other way around.
A repository will then be represented through its interface inside the domain but have its concrete implementation outside the domain, in the persistence module. This way the domain depends only on the abstract interface, not the concrete implementation.
That basically is object-orientation versus procedural programming on an architectural level.
See also the Ports and Adapters a.k.a. Hexagonal Architecture.
Another advantage of the repository is that you can create similar access mechanisms to various data sources. Not only to databases but to cloud-based stores, external APIs, third-party applications, etc.
You're right,in those simple cases the repository is just another name for a DAO and it brings only one value: the fact that you can switch EF to another data access technique. Today you're using MSSQL, tomorrow you'll want a cloud storage. OR using a micro orm instead of EF or switching from MSSQL to MySql.
In all those cases it's good that you use a repository, as the rest of the app won't care about what storage you're using now.
There's also the limited case where you get information from multiple sources (db + file system), a repo will act as the facade, but it's still a another name for a DAO.
A 'real' repository is valid only when you're dealing with domain/business objects, for data centric apps which won't change storage, the ORM alone is enough.
It would be useful in situations where you have multiple data sources, and want to access them using a consistent coding strategy.
For example, you may have multiple EF data models, and some data accessed using traditional ADO.NET with stored procs, and some data accessed using a 3rd party API, and some accessed from an Access database living on a Windows NT4 server sitting under a blanket of dust in your broom closet.
You may not want your business or front-end layers to care about where the data is coming from, so you build a generic repository pattern to access "data", rather than to access "Entity Framework data".
In this scenario, your actual repository implementations will be different from each other, but the code that calls them wouldn't know the difference.
Given your scenario, I would simply opt for a set of interfaces that represent what data structures (your Domain Models) need to be returned from your data layer. Your implementation can then be a mixture of EF, Raw ADO.Net or any other type of Data Store/Provider. The key strategy here is that the implementation is abstracted away from the immediate consumer - your Domain layer. This is useful when you want to unit test your domain objects and, in less common situations - change your data provider / database platform altogether.
You should, if you havent already, consider using an IOC container as they make loose coupling of your solution very easy by way of Dependency Injection. There are many available, personally i prefer Ninject.
The domain layer should encapsulate all of your business logic - the rules and requirements of the problem domain, and can be consumed directly by your MVC3 web application. In certain situations it makes sense to introduce a services layer that sits above the domain layer, but this is not always necessary, and can be overkill for straightforward web applications.
Another thing to consider is that even when you know that you will be working with a single data store it still might make sense to create a repository abstraction. The reason is that there might be a function that your application needs that your ORM du jour either does badly (performance), not at all, or you just don't know how to make the ORM bend to your needs.
If you are wrapping your ORM behind a well thought out repository interface, you can easily switch between different technologies as you see fit. It's not uncommon in my repositories to see some methods use EF for their work and others to use something like PetaPoco, or (gasp) ADO.net code. The repository abstraction enables you to use exactly the right tool for the job at hand without leaking these complexities into the client code.
I think there is a big misunderstanding of what many articles call "repository." And that's why there are doubts about what real value those abstractions bring.
In my opinion the repository in it's pure form is IEnumerable, while you and many articles are talking about "data access service."
I've blogged about it here.
I current got a flow like this:
The Repository gets injected with a Datasource. The ViewModel gets injected with the Repository.
Because there's a constant flow of Items and Mutations (deletions, changes) that need to translate to the UI I don't keep a collection of items in the Repository. Therefor I ended up with a repository that only passes along Items between the Datasource and the ViewModels.
I've always understood that you shouldn't directly use the Datasource in the UI layer (ViewModels are in my UI layer). But is this still the case if you use dependency injection? Am I putting the Repository pattern to its use or is it just causing me overhead? (Right now it feels like it does.)
I originally implemented the repository to decouple the UI and Datasource but I've found that DI does a great job without a Repository.
You might want to have a look at some of the blog posts by Oren Eini (aka Ayende Rahien)
The evils of the repository abstraction layer
or
Repository is the new singleton
More often than not repositories are just "flow heaters". They do nothing but forward calls and results and just add overhead.
Btw: if you should or should not use repositories has nothing to do with dependency injection!
Update
Oren just published a new post on this topic.
I am developing a WPF desktop app that uses Entity Framework 4 and SQL Compact 4. I have seen two distinct styles of Repository classes:
The Repository instantiates an ObjectContext, which is disposed of when the Repository is garbage-collected. The lifetime of the ObjectContext is the same as the lifetime of the application.
A separate DataStoreManager class creates and holds an ObjectContext for the life of the application. When a repository is needed, a command gets an ObjectContext reference from the DataStoreManager and passes it to the constructor for the New Repository. The lifetime of the ObjectContext is the lifetime of the application.
Is either approach considered a bad practice? Does either present any absolute advantages over the other? Is either approach considered best practice? Is either more widely accepted or used by developers than the other? Thanks for your help.
I would have thought that holding an ObjectContext open over multiple accesses would be bad practice. As soon as it becomes corrupted then you would need to recycle in and handle the corruption.
The Repository pattern is more for abstraction of data access but doesn't necessarily map to lifetime of context.
The unit of work pattern is more about encapsulation of one or more database/repository accesses i.e. a use case may have you adding a new Blog and then adding the first default post, this may require calling two repositories, at this point you might want to share the context and encapsulate these two commands in a transaction. Adding a second post might be done hours later and be a new context/unit of work.
DJ is right in mentioning Context lifetimes which you'd set generally at an application level.
The best practice is depended on how your users are going to use the application:
And how your application is structured.
if there is only one user using your application at one time, you can even create your entity context as a static instance.
context can be used per request, per thread, per form.
read more: http://blogs.microsoft.co.il/blogs/gilf/archive/2010/02/07/entity-framework-context-lifetime-best-practices.aspx