An introduction to GraphQL and Apollo

GraphQL is a query language and its big advantage is its flexibility. It allows a client to define a single query specifying exactly what data it requires. The back-end will then return a JSON object containing exactly this data, in precisely the same nested structure as the query. The benefits of this are huge.

Gone are the days of having to call multiple REST endpoints to get the data for a single screen, and ending up with a ton of surplus data that is never used. Instead, a single call will get all the data required for the screen. This solves the problem commonly known as overfetching.

Consider an instance where a client wishes to change their requirements for a particular screen. Say they wish to add an additional list to the screen which is populated with data from a server. With a REST API approach, the back-end dev would have to change an existing endpoint or create a new endpoint to feed this data.

The front-end dev would then have to call this endpoint, parse the data and then display it on the screen - quite a bit of work for a relatively small change. In contrast, with a GraphQL approach, the back-end dev would need to expose the new data to the existing endpoint. The front-end dev would then just need to change the existing query to fetch the data. Far more efficient and far more flexible.

GraphQL is also pretty straightforward to learn. And with a rich ecosystem developing, there is excellent tooling, documentation and tutorials available. Given companies like Facebook, GitHub, Shopify and Twitter all use GraphQL, these resources are only likely to continue growing.

Whilst GraphQL has many advantages, like anything, it also has a few disadvantages.

The first of which is rate limiting. Given that a client can specify the exact data it requires, a GraphQL operation can be of varying load, anywhere from cheap to expensive. Therefore, unlike a REST API where you can just specify how many times an endpoint can be hit per day, it’s much more difficult to rate-limit a GraphQL backend.

The second is that you need to learn how to use GraphQL. Whilst this may seem like an obvious thing, there is always a time-cost debt in learning new technologies.

The third is that given queries are sent from the client, we need to use some sort of client library to handle the GraphQL libraries. This results in extra code in the client.

One of the most popular and widely used client libraries is Apollo. Apollo is a networking library that makes it easy to consume any GraphQL schema, fetch and mutate data from any GraphQL endpoint and even auto-generate data models for you. It has great documentation, and code examples, and is supported on many different platforms including native iOS, native Android and JavaScript frameworks such as React, Angular and Vue.

We have only used the Apollo Framework for iOS and have overall been very impressed with it. Whilst I experienced a few issues with the initial setup, my experience since then has been really positive and it has been great to use.

Another awesome tool is GraphQL Playground. This is the equivalent of Postman for a REST API and allows you to test out an endpoint, verify the responses and also view the schema. This is a must-have if you’re working with a GraphQL backend.

Image source.

In summary, GraphQL is a really cool, powerful technology that can be used to power your apps. If you haven’t already done so, we’d strongly recommend you check it out, play around with it and see for yourself the awesome benefits it offers. We reckon it will be powering our mobile apps for some time to come.

The new version of our backend platform The Core is designed around AWS microservices and GraphQL. For more info or advice on GraphQl feel free to contact us.

Useful Links

[1] GraphQL - https://graphql.org

[2] GraphQL vs REST - https://blog.apollographql.com/graphql-vs-rest-5d425123e34b

[3] GraphQL and Apollo on iOS - https://www.raywenderlich.com/595-getting-started-with-graphql-apollo-on-ios

[4] Apollo - https://www.apollographql.com