GraphQL in e-commerce: Building for the next billion users

This talk highlights the benefits of building scalable applications using GraphQL, the challenges that come with it and how you can better manage these as your application scales to an increasing number of users.

Transcript

1. GraphQL in e-commerce: Building for the next billion users Fredrick Mgbeoma Tuesday, 29. September 2020 Tweet: @fuchodeveloper #GraphQLBerlin
2. About Fredrick • Software Engineer • Blog - codeisbae.com • Twitter - @fuchodeveloper • GitHub -github.com/fuchodeveloper • Web accessibility advocate • Open Source lover Tweet: @fuchodeveloper #GraphQLBerlin
3. Why GraphQL at commercetools
4. API-Driven
5. • 2012 - Created by Facebook • 2015 - Released to the public • 2016 - Adopted at commercetools. GraphQL API made public • 2018 - 33% coverage at commercetools • 2020 - 81% coverage of the core commerce APIs at commercetools • Present - Moving to achieve 100% coverage* * commercetools has the highest GraphQL coverage in the e-commerce industry with a goal to reach 100%
6. GraphQL at commercetools
7. Modern e-commerce shop
8. Performance • Reduced network usage • Consumes less resources for your users • Great when building for mobile clients
9. Speed • Huge speed gains low-end devices and wearable • Goodbye to unnecessary payloads • Improved overall user experience
10. New products Average loading time reduced from 3100ms to 1000ms Apple Watch Series 2, loading new products using REST vs GraphQL API
11. How does GraphQL make this possible?
12. Ask for what you need, get exactly that • Describe your data
13. Ask for what you need, get exactly that • Request only the data you need
14. Ask for what you need, get exactly that • Get predictable results
15. A billion requests
16. Contentful: Over a billion GraphQL requests per month Source: https://web.facebook.com/contentful/photos/a.663496390331596/3812957512052119
17. Facebook: About 260 billion requests per day - Nick Schrock Source: https://www.b2bnn.com/2015/09/how-facebook-moved-from-the-desktop-to-mobile-with-graphql/ https://twitter.com/reactnewsletter/status/635909339748798464
18. Challenges building at scale with GraphQL
19. Complex queries • Performance issues for deeply nested queries
21. Database operations • A single query can potentially generate a lot of database calls
22. How we solved them
23. Sangria
24. Amount limiting • Error out if query limit is above 500
25. Optimisations • Default limit for results returned for API calls
26. Pre-computed queries • Proper validation of queries by client based on the schema
27. Cost calculator • Compute how expensive requests are and if not performant, it is rejected • Complexity limit of 20000 • Sangria default complexity, can be customised per field
28. Request
29. Error response
30. How do I analyse the cost of my queries?
31. Magic
32. 01 Inspecting the x-graphql-query-complexity response header and its value.
33. 02 Use the GraphiQL’s Profiling functionality to measure the impact
34. Demo using Impex
35. Impex link
36. Resources • https:/ /medium.com/@oleg.ilyenko/profiling-graphql-6733 0511f09c • https:/ /www.apollographql.com/blog/securing-your-graph ql-api-from-malicious-queries-16130a324a6b/ • https:/ /docs.github.com/en/free-pro-team@latest/graphql /overview/resource-limitations • https:/ /techblog.commercetools.com/building-an-ecomme rce-graphql-api-the-challenges-6d652a95f478
37. Thank you.