GraphQL best practices for high performance data fetching from database?

Hello,

How can I improve performance of GraphQL queries? This is a general question, I know. I’m looking for best practices on how to implement GraphQL with graphql-java.

I have a simple schema:

type Account @java(package:"com.example") {
  id: ID!
  users: [User]!
}

type User @java(package:"com.example"){
  id: ID!
  email: String!
  firstName: String
  lastName: String
}

Making the following query:

{
  accounts {   id   users {   id  }   }
}

takes around ~19s -> ~50s on my machine and the staging server to produce: query-results.txt

Data is fetched from a database with a code that looks like this:

    @Override
    public List<Account> getAccounts() {
        List<gr8pi.web.platform.Account> allAccounts = accountRepository.findAll();
        return allAccounts.stream().map(a -> {
                List<gr8pi.web.candidate.Respondent> respondents = respondentRepository.findAll();

                List<Membership> memberships = membershipRepository.findByAccountId(a.getId());

                List<User> users = memberships.stream()
                    .map(membership -> new User.Unresolved.Builder()
                        .withId(Long.toString(membership.getUser().getId()))
                        .withEmail(membership.getUser().getEmail())
                        .withFirstName(membership.getUser().getFirstName())
                        .withLastName(membership.getUser().getLastName())
                        .build())
                    .collect(Collectors.toList());

                return new Account.Builder()
                    .withId(Long.toString(a.getId()))
                    .withRespondents(Collections.EMPTY_LIST)
                    .withUsers(users)
                    .build();
            }
        ).collect(Collectors.toList());
    }

I’m using spring data repository and hibernate 5.2 and the database is PostgreSQL. The code is very simple and I do believe some improvements can be made, but the point is, in my opinion : you can’t get away without making a lot of calls to the database.

My question is: What are the best practices on implementing good performing data fetching from database systems.

Thanks,

[bbakerman]

Also have a look at idea behind DataLoader : https://github.com/facebook/dataloader

It allows for combining common look ups into less calls (if it suits your app)

There is a java implementation : https://github.com/engagingspaces/vertx-dataloader

I havent used the above one however

[bbakerman]

In terms of pagination in graphql, this is your responsibility as a schema designer.

The list fields you give back can contain “pagination” arguments and pagination “info”

The Relay specification has pagination built in : https://facebook.github.io/relay/docs/graphql-connections.html

This also outlines some ways to do pagination : http://graphql.org/learn/pagination/

The point is pagination is not a graphql engine thing - its the schema designers repsonsibility