The way you are doing DI is wrong.
First, the connection string belongs in the data layer. Or in the web.config file.
The next abstraction you will be dealing with is the DbContext, not a connection string. Your repositories should not know about connection strings. Your business logic will not know about DbContext etc.
Your UI will have no idea and will not instantiate anything related to EF.
Concrete answers to your points:
Do not add abstractions, until you are very familiar with EF. It already adds good abstractions like UoW, queries, using POCOs etc.
For DI to work, you have a Composition Root which references all components needed. This might or might not be in the WebUI project. If it isn't, you should expect that it does not reference EF or any other data related tech.
Stop right here. Stop adding abstractions over abstractions. Start with direct and 'naive' architecture and develop it over time.
Abstractions are a tool to deal with complexity. Absence of complexity means no abstractions needed (yet).
Your case
In your case all three options are viable. I think that the best option is probably to sync your data sources someplace the asp.net application is not even aware of. That is, avoid the two fetches in the foreground every time, sync the API with the db silently). So if that's a viable option in your case - I say do it.
A solution where you make the fetch 'once' like the other answer suggests doesn't seem very viable since it doesn't persist the response anywhere and ASP.NET MVC will just make the fetch for every request over and over.
I'd avoid the singleton, I don't think it's a good idea at all for plenty of the usual reasons.
If the third option is not viable - one option is to lazy load it . That is, have a class extend the entity, and have it hit the API on a need to basis. That's a very dangerous abstraction though since it's even more magic and non-obvious state.
I guess it really boils down to several questions:
- How often does the API call data change? Not often? Third option. Often? Suddenly the third option is not too viable. I'm not sure I'm as against ad-hoc calls as you.
- How expensive is an API call? Do you pay per call? Are they fast? Free? If they're fast, making a call each time might work, if they're slow you need to have some sort of prediction in place and make the calls. If they cost money - that's a big incentive for caching.
- How fast does the response time have to be? Obviously faster is better, but sacrificing speed for simplicity might be worth it in some cases, especially if it's not directly facing a user.
- How different is the API data from your data? Are they two conceptually different things? If so, it might be even better to just expose the API outside rather than return the API result with the result directly and let the other side make the second call and handle managing it.
A word or two about separation of concerns
Allow me to argue against what Bobson is saying about separation of concerns here. At the end of the day - putting that logic in the entities like that violates separation of concerns just as bad.
Having such a repository violates separation of concerns just as bad by putting presentation centric logic in the business logic layer. Your repository is now suddenly aware of the presentation related things like how you display the user in your asp.net mvc controllers.
In this related question I've asked about accessing entities directly from a controller. Allow me to quote one of the answers there:
"Welcome to BigPizza, the custom Pizza shop, may I take your order?" "Well, I'd like to have a Pizza with olives, but tomato sauce on top and cheese at the bottom and bake it in the oven for 90 minutes until it's black and hard like a flat rock of granite." "OK, Sir, custom Pizzas are our profession, we'll make it."
The cashier goes to the kitchen. "There is a psycho at the counter, he wants to have a Pizza with... it's a rock of granite with ... wait ... we need to have a name first", he tells the cook.
"No!", the cook screams, "not again! You know we tried that already." He takes a stack of paper with 400 pages, "here we have rock of granite from 2005, but... it didn't have olives, but paprica instead... or here is top tomato ... but the customer wanted it baked only half a minute." "Maybe we should call it TopTomatoGraniteRockSpecial?" "But it doesn't take the cheese at the bottom into account..." The cashier: "That's what Special is supposed to express." "But having the Pizza rock formed like a pyramid would be special as well", the cook replies. "Hmmm ... it is difficult...", the desparate cashier says.
"IS MY PIZZA ALREADY IN THE OVEN?", suddenly it shouts through the kitchen door. "Let's stop this discussion, just tell me how to make this Pizza, we are not going to have such a Pizza a second time", the cook decides. "OK, it's a Pizza with olives, but tomato sauce on top and cheese at the bottom and bake it in the oven for 90 minutes until it's black and hard like a flat rock of granite."
(Read the rest of the answer, it's really nice imo).
It's naive to ignore the fact there is a database - there is a database, and no matter how hard you want to abstract that, it's not going anywhere. Your application will be aware of the data source. You won't be able to 'hot swap it'. ORMs are useful but they leak because of how complicated the problem they solve is and for plenty of performance reasons (Like Select n+1 for example).
Best Answer
Yes.
Here's why: you'll need a common, web-enabled API for your three front-ends to talk to. Entity Framework won't execute queries to your DB server over the Internet.
No. You can use it to build the Web API. You can even retain your Repository/Unit of Work layers, if you want to. Here's what your architecture might conceptually look like: