Apollo server lambda architecture

I’m working on the architecture for a new project, the general idea is that it should be a graphql server with apollo server, typescript and we will be using AWS cloud services.

I was thinking of creating a serverless microservice architecture using apollo server lambda, getting an EC2 to set up the apollo federation, and then creating all functionality using lambda functions, unfortunately, there is no good documentation on the topic, and I would like to know if someone here has experience with this kind of architecture and can help to answer some questions.

I plan to create a repository per microservice and in there create all the lambda functions (I mean create functionality according to the use case per lambda), but according to the npmjs package it should be one apollo server per lambda function, with this approach I will end having a thousand servers, I don’t think that’s the best approach, do you have any recommendation on this?

I’m attaching a simple architecture diagram and a simple POC that I ran just a second ago GitHub - wilvermelendez/apollo-server-lambda: This a POC using apollo server lambda with typescript (In the repository I have two entities worker and journey probably but I wanted to test if I could create 2 lambda functions with one apollo server but it seems that it does not work in that way)

I too am very interested in this topic. I recently posted a Stack Overflow on this question. I like your diagram because it keeps everything within AWS, i.e. not going out to the internet and back in to create sub graphs. I presume all the sub graph apis are private? Interested in a follow up if you found out anything on this.