Dataloader

DataLoader is a generic utility to be used as part of your application’s data fetching layer to provide a simplified and consistent API over various remote data sources such as databases or web services via batching and caching. It is provided by a separate package aiodataloader <https://pypi.org/project/aiodataloader/>.

Batching

Batching is not an advanced feature, it’s DataLoader’s primary feature. Create loaders by providing a batch loading function.

from aiodataloader import DataLoader

class UserLoader(DataLoader):
    async def batch_load_fn(self, keys):
        # Here we call a function to return a user for each key in keys
        return [get_user(id=key) for key in keys]

A batch loading async function accepts a list of keys, and returns a list of values.

DataLoader will coalesce all individual loads which occur within a single frame of execution (executed once the wrapping event loop is resolved) and then call your batch function with all requested keys.

user_loader = UserLoader()

user1 = await user_loader.load(1)
user1_best_friend = await user_loader.load(user1.best_friend_id))

user2 = await user_loader.load(2)
user2_best_friend = await user_loader.load(user2.best_friend_id))

A naive application may have issued four round-trips to a backend for the required information, but with DataLoader this application will make at most two.

Note that loaded values are one-to-one with the keys and must have the same order. This means that if you load all values from a single query, you must make sure that you then order the query result for the results to match the keys:

class UserLoader(DataLoader):
    async def batch_load_fn(self, keys):
        users = {user.id: user for user in User.objects.filter(id__in=keys)}
        return [users.get(user_id) for user_id in keys]

DataLoader allows you to decouple unrelated parts of your application without sacrificing the performance of batch data-loading. While the loader presents an API that loads individual values, all concurrent requests will be coalesced and presented to your batch loading function. This allows your application to safely distribute data fetching requirements throughout your application and maintain minimal outgoing data requests.

Using with Graphene

DataLoader pairs nicely well with Graphene/GraphQL. GraphQL fields are designed to be stand-alone functions. Without a caching or batching mechanism, it’s easy for a naive GraphQL server to issue new database requests each time a field is resolved.

Consider the following GraphQL request:

{
  me {
    name
    bestFriend {
      name
    }
    friends(first: 5) {
      name
      bestFriend {
        name
      }
    }
  }
}

If me, bestFriend and friends each need to send a request to the backend, there could be at most 13 database requests!

When using DataLoader, we could define the User type using our previous example with leaner code and at most 4 database requests, and possibly fewer if there are cache hits.

class User(graphene.ObjectType):
    name = graphene.String()
    best_friend = graphene.Field(lambda: User)
    friends = graphene.List(lambda: User)

    async def resolve_best_friend(root, info):
        return await user_loader.load(root.best_friend_id)

    async def resolve_friends(root, info):
        return await user_loader.load_many(root.friend_ids)