I have a set of Django microservices, each exposing GraphQL APIs. Currently, I'm using Apollo Server with Apollo Federation to combine all subgraphs into a single supergraph. However, I've encountered a few limitations:
Some of Apollo’s advanced features aren't free.
The rover command (used to create a supergraph from subgraphs) requires an internet connection, which isn't ideal for my setup.
I'm looking to build a custom solution in Python (using Flask, Django, or similar) that can serve as both an API gateway and a GraphQL federation service. The goal is to aggregate all subgraphs and generate a supergraph without relying on Apollo's proprietary tools.
Has anyone implemented a similar solution or have suggestions for frameworks, libraries, or strategies to accomplish this? Any advice on building this gateway and handling GraphQL federation in Python would be greatly appreciated.
I'm looking to build a custom solution in Python (using Flask, Django, or similar)
There are a few alternative GraphQL gateways (free) nowadays, check the audit I did to measure how compatible they are with Apollo Federation spec: https://the-guild.dev/graphql/hive/federation-gateway-audit
I would say it's hard to build your own gateway, because of the complexity of the Apollo Federation spec and there are literally 0 libraries that will help you build one. When I say it's complex, I mean it, just look at the tests I created for the audit (it's open source).
You could potentially use Apollo's query planner (JavaScript, but soon also in Rust) to generate a Query Plan, that you will execute in Python, but is it worth the effort when you have so many options that have good plugin system?