Ask HN: How has GraphQL panned out for your organization?

We're using GraphQL at my startup, a bootstrapped unicorn company that is, for lack of a better description, the Uber of Rape.

It's been great, except that one time when I was testing, I sent a buggy query to our production instance, and it resulted in the CEO's mother being raped to death. Sometimes, when I go to sleep at night, I can hear her screaming. Sad stuff, I wish that didn't happen.

It's worked out great for us. We use it as a flexible way to get exactly what is needed in a single request that ends up pulling data from a mongo database. It has not added any extra complexity. On the contrary it has simplified out backend and front end apps. We use go and https://github.com/UHN/ggql.

I didn't choose to adopt GraphQL, but I joined a company where an ex-developer had integrated into what was at that point a core codebase. They didn't add it due to any particular merit, they just wanted to play with shiny new technology. The codebase also had REST endpoints, but GraphQL made up the majority of the API, so we decided to just run with it and deprecate REST. At the time the decision we made was arbitrary.

This turned out in retrospect to be a good call. The primary consumers of the API were a set of Android and iOS apps, and integrating GraphQL into mobile apps is a pleasant experience. You can generate your data models from the graph schema. Tools like GraphiQL allow developers to explore and play with your API, substantially reducing the amount of documentation needed. The mobile team much preferred working with GraphQL over REST.

On the backend, GraphQL has a bit of a learning curve. The resolver pattern is powerful once you understand how to use it properly, but also easy to misuse. The most common source of problems is developers not understanding Graph's batch resolution mechanism. Developers accustomed to building REST APIs rely on preloading data, which is the wrong approach when dealing with GraphQL.

As a simple example, maybe you initially build a Graph query where you load comments like this:

  query Comments {
    user {
      posts {
        comments {
          message
        }
      }
    }
  }

But later on, you decide you want to allow a user to load their comment history directly:

  query Comments {
    user {
      comments {
        message
      }
    }
  }

In the first example, an equivalent REST API might load comments via something like posts.preload(:comments), and when the second query is built you'd use user.preload(:comments). But the way to solve this problem in Graph is batch resolution - get a list of comment IDs you want to load, then call Comment.where(id: comments_ids). Your comment loading is now context agnostic.

Another related problem is that it is easy to create N+1 queries. This also stems from a misunderstanding of batch resolution - reusing the above example it would be the equivalent of calling Comment.find(comment_id) for each ID instead of Comment.where(id: comments_ids). My first few months working with Graph were often spent fixing N+1 queries. As much as I try to explain this problem to new developers, it's one of those things that doesn't click for them until they cause the problem and end up having to fix it.

Once you understand GraphQL it's a nice technology to work with. In the early stages it can be fairly easy to shoot yourself in the foot, you just need to build up enough knowledge to avoid the common pitfalls. In general, GraphQL APIs are harder to build than REST APIs (mostly because REST is the default for most frameworks). If I had to start a new project today, my decision on whether to use GraphQL would be be determined by questions such as:

  1. How much control do I have over the clients
  2. Are the primary consumers native mobile applications 
  3. Is flexibility important
  4. Will API documentation be required

I have been using graphql since 2019 and it has been great for me. Overall if you have a big team, it is much more easier to manage. The graphql ecosystem is growing and tooling are standadized (nowadays the N+1 can be easily solved)

But one thing will always remain and you have to remember is that graphql is slow by design. If performance is critical to you, don't use it

I currently work on a project using GraphQL from the start and we are happy with this choice. It's sometimes a bit complex but when I listen to some friends debating about how to build the most beautiful RESTful API for their use case, I don't regret GraphQL at all.

We have some queries that are a bit slow because it's easy to ask a lot in GraphQL, but the alternative would be to do hundreds of HTTP queries if we were using REST for example.

The tooling is particularly good nowadays. My company use it in NodeJS with Apollo and TypeScript and it's very good and stable. I also did some side projects in Rust and it was very pleasing to use too.

One common problem is the caching, you can't easily cache using normal HTTP CDN software for example, but since our API requires to be authenticated for anything, it's not a problem for us. We just do cache in other places.

We have some simpler API not using GraphQL but I would strongly recommend GraphQL as soon as your API is not tiny.

At a minimum, GraphQL skips the problem of having one or few engineers re-invent the wheel of how to pass queries to an API. Or if you’re lucky, how to document it.

REST Api design is probably one of the least value-add activities a company can involve itself in. Any standard that defines a standard way of performing set operations, GraphQL or even the ancient ODATA, is superior.

You're likely to get a lot of selection bias in the answers - people using it likely made a decision that it's worth the hassle for their use case. To add a counter point: GraphQL was a non starter for our company, because we work with _very_ heterogenous clients who need to implement our API and the extra learning curve would be a killer. If you have to implement a simple REST API you just look at it and in a few hours the job is done. If it would be GraphQL, lots of people would have to learn a new paradigm just to implement a few calls.

I used it to built my startup (currently bootstrapping, just me) and I have no regrets. The data model is relatively complex and the frontend makes very heavy use of Apollo features. I think the number one benefit in the backend is just having a well documented data model that isn’t tied to specific (arbitrary routes). For the frontend, using Apollo makes things like consistency very easy in a SPA since it normalizes all the data it receive (there’s no “data tearing” that tends to come when you load initial data from source A then update from source B). Also, using codegen for the frontend is chefs kiss (I hate manually creating TypeScript interfaces for API response shape, especially since they can get out of date easily).