Let's say we have a simple food delivery app. Where client order the food, then restaurant start preparing the food and gives it to the courier who delivery it to the client.
So here we have three different domains and each of this domain have their own order:
client - here client order the food and have the status of the food in preparation | in delivery | delivered
restaurant - here restaurant got its order and has their own status in queue | in preparation | ready to pick up
courier - courier has only two status delivering | delivered
Moreover each of this domain has their own price and other attribute about order:
client - total price (food price + delivery cost + fee)
restaurant - price of food, time of production to give a hind to the client when food will be delivery
courier - cost of delivery
All I want to highlight is that each of the domain has its own order aggregate, so according to DDD we have to keep it in different aggregates even in different microservices:
client - /orders/:id provides the general status of the order and total price to the client.
restaurant - /restaurants/:restaurantId/orders/:id provides the status of the food in restaurant domain and cost.
courier - /couriers/:courierId/orders/:id provides information how much courier earn from this order and how long it took to delivier
But now I met another problem, because client order combines information from other domains (is food still in restaurant or it's being delivery) so I have to gather this information when client asks about its order, but it means that client doesn't have its domain (its own aggregate, total price, discount etc), but if I create order aggregate for the client then I will not keep all information about order in one place (when restaurant give the food to the courier it should also change status of the order in client domain) what is not really according to microservices, because we keep information about the same order in different microservices.
Should I just create one order domain or should I split it to different domain and make these domains communicate between, when something will change in one domain?
One useful approach is to leverage domain events. When the restaurant's view of the state of the order changes, an event describing that change is published. The other services can then update their model of the event (assuming that that change is relevant to that service).
So for instance, we might have:
user creates order via the client service => OrderCreated event emitted
restaurant service consumes OrderCreated event, translates the order for the restaurant (e.g. uses the prices which the delivery app pays the restaurant vs. the prices the delivery app charges the user) => OrderSentToRestaurant event emitted
courier service consumes OrderCreated and begins trying to figure out which courier will be assigned the order and the approximate transport time from pickup to delivery => DeliveryLatencyEstimateMade event emitted
client service consumes OrderSentToRestaurant and updates its order status (for presentation to the user) to in preparation
courier service ignores OrderSentToRestaurant
restaurant service ignores DeliveryLatencyEstimateMade event
client service consumes DeliveryEstimateLatencyEstimateMade and updates its model (delivery time remains unknown)
restaurant informs restaurant service of expected completion time => OrderReadyForPickupAt event emitted
courier service consumes OrderReadyForPickup, refines courier assignment decisions
client service consumes OrderReadyForPickupAt event, combines with the latest latency estimate to present a predicted delivery time to the user
and so forth. Each service is autonomous and in control of its data representation and free to ignore or interpret the events as it sees fit. Note that this implies eventual consistency (the restaurant service will know about when the order is expected to be ready for pickup before the courier or client services know about that), though microservice autonomy already effectively ruled out strong consistency.
When looking at aggregate design in each bounded context (BC), you have to include only the data required to provide the functionality that belongs to that BC. The fact that the restaurant endpoint needs to return some extra data is not a good enough reason to add that data to the order aggregate in that BC.
You can resolve the need for more data in different ways:
The API client can call multiple endpoints to fetch all the data it needs
The API can implement Data Aggregation, by internally querying multiple BCs/microservices and combining them to produce a single more complete response object
Create Read models, which store data from multiple sources into a single "table" in a way that simplifies querying and returning this data. This approach is more complex, but it's very useful when you need to filter and sort by fields belonging to multiple BCs, which is not possible with the previous two approaches.
Another consideration to make is double-checking if your boundaries are correct. Do you really need a Client BC? What business logic does it implement? Maybe Orders are created directly into Restaurant and there is no Client order? Client order could just be a "façade" providing all Restaurant orders belonging to a single client Id?
As a final note, I completely agree with Levi Ramsey's answer that events are the right way to coordinate the different aggregates. They would also be used to create the read models I mentioned above.
Related
to give you a bit of context, I'm developing a game, an online soccer manager, and I have the following microservices:
Clubs
Season
The Clubs microservice takes care of the club management and the Season is responsible for the Season management.
One of the responsibilities of the Season service is return the league standings, with the club names and their positions. In the Season service, I only store the club_id, but to fulfill the request to return the standings, I would need also the club name, which resides in the Clubs service.
Now, I could implement a REST endpoint in the Clubs service to return the club name, but them those service won't be loosely coupled anymore.
As I saw from my readings, I have 2 options, and they are:
Have a clubs cache in the Season service, where it does the relationship between the club_id and club_name (Could be a database table). In this case the data will be duplicated (which is OK for most of the cases), but I need to keep in sync with the domain events dispatched by the Clubs Service.
The other option would be create another microservice to be used as API composition pattern. So this API would get data from both service, enrich the response and send back to the caller.
Now, I'm in doubt which approach should be taken. Which one has less downsides?
Both are described in Saga pattern. There are pros and cons of both. You have to choose based on your NFRs.
Let’s assume I’m in the context of an admin panel for a webshop. I have a list of orders. Those orders are payed for and are ready to ship. The (admin) user would like to start making shipments based on the items ordered.
Imagine there are 2 microservices. One for orders and one for shipments. In order to create a shipment, i will send a request with a couple of items to be shipped and an order ID to the shipment service. The shipment service will then check whether the items are present in the order by querying the order service. Because i don’t want to create a shipment with items that are not present in the order.
I’d like to have immediate consistency because the shipment data will be send to a third-party application after creation. Thereby it also feels weird to allow shipments to be created if the data is not correct.
I’m also using GraphQL mutations. Which means i have to return the updated state to the user, which also makes eventual consistency a lot harder.
What is the recommended approach for these situations? Could this be a sign that these 2 microservices need to be merged? I can imagine this situation can occur multiple times.
Suppose we have two microservices, Customers and Orders, with no dependencies between them, i.e. they don't call each other, they don't know each other. Every order, though, has a reference to a customer by means of a customer id. In other words one customer may have zero or more orders, and one order belongs to exactly one customer.
For the sake of the example, it's totally fine to delete a customer unless there are orders belonging to that customer. What is the most appropriate way to implement a constraint on Customers that prevents a customer from being deleted if one or more orders have a reference to that customer? Analogous to referential integrity in a relational database.
These are the approaches I can think of:
Let Customers ask Orders if a given customer has any orders, e.g. via API call.
Let Customers keep track of which orders are assigned to every customer, e.g. by having each customer record maintain a list of order ids.
Merge Customers and Orders into a single microservice and manage it internally.
I'm not sure which approach is the best for the given example in a microservices context. I can see pros and cons in all three approaches but I won't list them here because I'd like to hear other people's thoughts on the problem, including approaches not listed above. Thanks.
Probably the second approach would help if you're going to decouple through events, either tracking a list of ids or a counter just telling how many orders are stored for such a Customer.
On the Order microservice you will emit an event when there is a creation/deletion that will be captured by the Customer (or any other microservice interested) who will take care of updating the list of order ids (or increment/reduce the counter).
If customer order counter is 0 then you may delete the customer.
Let's start with your third approach: This will not work in a Microservice world, because you will always have those constraints between some Services. And if you want to solve all of them this way, you'll end up with a Monolith - and that's the end of your Microservice story.
The first and second approach have both the same "problem": These are async operations, that may return false positive (or false negative) results: It's possible to make api requests for delete customer and create order (or delete order) at the same time.
Though this can happen:
For your first approach: Customer Service asks Order Service if there are any Orders for this Customer. Order Service returns 0. And at the same time Order Service creates a new Order for that Customer in another thread. So you end up with a deleted Customer and still created an Order.
The same applies for your second approach: The messaging between those services is async. Though it's possible that Customer Service knows of 0 Orders, and permits the delete. But at the same time the Order Service creates a new Order for this Customer. And the OrderCreated message will hit the Customer Service after the Customer has already been deleted.
If you try to do it the other way around, you'll end up with the same situation: Your Order Service could listen to CustomerDeleted messages, and then disallow creating new Orders for this Customer. But again: This message can arrive while there are still Orders in the database for this Customer.
Of course this is very unlikely to happen, but it still is possible and you cannot prevent it in an async Microservice world without transactions (which of course you want to avoid).
You should better ask yourself: How should the system handle Orders where the corresponding Customer has been deleted?
The answer to this question is most likely dependent on your business rules. For example: If the Order Service receives a CustomerDeleted message, it may be okay to simply delete all Orders from this Customer. Or maybe the behavior depends on the Order's state property: It's okay to delete all Orders with state = draft, but every other Order from this Customer should still be processed and shipped as usual.
In the context of a Microservice architecture, a single business operation can require collaboration between two or more services.
Suppose we have an Order Management Service and a Product Catalog Service.
When the user adds an order item to an order, the Order Management Service will persist a OrderItem object which have the following attributes among many others :
OrderItem
+ Id
+ ProductId
+ ProductName
In order for the Order Management Service to fill the ProductName attribute, we have 4 choices as I see it :
Choice 1 : ProductName is given by the client app as it probably already has this data from previous requests
Choice 2 : If the architecture uses an Api Gateway, the gateway will be responsible for retrieving the ProductName from the Product Catalog Service then provide it to the Order Management Service.
Choice 3 : The Order Management Service will call directly the Product Catalog Service and asks him for the ProductName givent a product id.
Choice 4 : The Order Management Service has a duplicate (but not exhaustive) product informations in its own database and these datas will be updated each time an update event is received from the Product Catalog Service.
Among these 4 choices, the n°1 seems not ok to me as we can't trust the client to give us a correct and updated value of ProductName.
I would like to have your opinion about what you think the best choice is and why !
Thanks !
Riana
Choice 1 : ProductName is given by the client app as it probably already has this data from previous requests
Like you said, it is not the best idea because the client may have stale information. Maybe acceptable if the product information changes infrequently and/or you have a secondary verification at order processing.
Choice 2 : If the architecture uses an Api Gateway, the gateway will be responsible for retrieving the ProductName from the Product Catalog Service then provide it to the Order Management Service.
IMHO, this is not a good idea. By doing so your domain/business logic will leak into the API Gateway. The gateway now knows the relationship between Orders and Products. This API gateway configuration/scripting will need to be maintained and introduces additional coupling.
Choice 3 : The Order Management Service will call directly the Product Catalog Service and asks him for the ProductName givent a product id.
This is my preferred choice. Though I do not recommend "direct" synchronous calls. Perhaps a retrieval of the ProductName via a messaging mechanism (message queue, event bus). Chained synchronous calls will reduce the availability of your services. You can find more information at Microservice Messaging Pattern.
Choice 4 : The Order Management Service has a duplicate (but not exhaustive) product informations in its own database and these datas will be updated each time an update event is received from the Product Catalog Service.
Data duplication is generally frowned upon unless there is a really good reason for it. In this case I don't see any. Why bother splitting the databases into two for each of the services yet duplicate the data between them? Also, to have the data updated each time an update event is received indicates that some kind of event/messaging infrastructure is available, in that case, why not just use messaging?
This duplication may be justifiable for high volume, low latency look ups, but it is a slippery slope that may end up with duplicated data all over your services. Imagine the repercussions of a length or type/format change of the ProductName string...
I am working on a SAAS project where users can create various projects. With each project, they can choose from 5 different plans. Each plan has its own costs per month. Hotjar is a kind of equal concept.
Now I want to arrange the subscription with Stripe. The problem with that was that a user can have a maximum x subscription, which of course was a shame. Then I decided to take 1 subscription that has several plans. But now I have a dilemma, to update the subscription you have to change the number via SubscriptionItem. Then you have to save yourself which plan has which SubscriptionItem_id for which user. That is quite a detour and can cause many problems.
Someone is a better way with Stripe or another payment software.
You don't necessarily need to store the subscritpion_item IDs, you can look it up via the subscription_item list API. All you need to do is store the subscription_id for your customers, and based on that ID you can retrieve the list of subscription_items:
\Stripe\Stripe::setApiKey("sk_test_9GavlLpfiKewqeCBXvRvmVgd");
\Stripe\SubscriptionItem::all(["subscription" => "sub_EQlPGjVj4o5luH"]);
Then you can handle the data part of the returned JSON object and update / delete / etc these subscription items.
If you only have the customer_id handy, then you can use the subscription list API (with status as well on the GET params) to retrieve the list of active subscriptions.