I have been improving the graphQL landscape at Keller Williams Realty International for the last 5 years. When I first started with the graph all I saw was how it had been done in weird ways that seemed bloated and inefficient. Before I pushed for us to rip it out and go back to rest I dug in to understand what problem was graphQL created to solve? I read some docs and watched this video on graphql. I learned that many of the things I hated about how I was seeing graphQL work was because of our implementation choices. I architected our transition from a monolith with many anti-patterns to a federation with only a few anti-patterns. I am a strong believer in the utility of making data available to our development teams using the graph. Even people in my company who have their reservations about moving everything to the graph can’t deny the utility when they are getting their hands dirty leveraging the connections.
We still have more services out of the graph than in but almost all of our most used interactions rely on the graph. looking forward to finding a way to keep the momentum rolling.