AWS API-GW integration with Realm Http service & GraphQL service

MongoDB Atlas App Services & Realm Atlas Triggers & Functions
,
1

Hi, We are thinking of using Realm. We’re are growing startup based in India.
Most of our backend services run in AWS VPC, and our APIs are exposed through AWS API-GW. I have some question regarding integrating Realm with AWS ecosystem -

  1. Can I expose Realm GraphQL endpoint and HTTP service through API GW ? The reason why we need this, is our clients who will be using our APIs. So it’s in best interest that domain name and url remains same for clients even though system behind them changes.

  2. As I have read, Realm HTTP service is used for low latency APIs, compared to using Lambda function with API GW for REST APIs, where cold start is measure issue. How does Realm manages cold start ? How is Realm function different than AWS Lambda function in terms of provisioned concurrency & pricing?

  3. Triggers have hard limitation of 3000 invocation per second which currently good enough number for us, Is there any option for increase in this quota if in future it requires more than that ?

  4. How does token resumability(streams) works let’s say for some reason AWS Eventbridge went offline or I change cluster’s configuration ?

  5. How to use caching service like Redis with Realm function ?!

Any help with these !!

2

Hi @Timey_AI_Chatbot

I will try to answer your questions.

  1. Realm provide a graphql rest web query see curl example :

Make sure to authenticate with a proper user and use the retrieved token

  1. MongoDB realm functions are running as an immediate execution they use context connection to access your data sources, those are optimised for MongoDB Access. I am not aware of any cold start with realm functions.

The cost is measured by various factors you can read more here

  1. The limitations is currently a given thing. If you can present a valid reason to increase please contact support to see if its possible.

  2. So resumabilty of a trigger is based on change stream resume tokens so it allows resuming a failed event listening on the database side as long as this event is present in MongoDB oplog.

If a function failed or aws event bridge the event will not resume on the particular event and only log an error its on the user to prepare a functionality to resume those for example reupdate all “undone” documents to be rerun by a trigger.

  1. I am not sure how those are connected, if a redis service cache realm function results for example than you only need to populate them. There is no direct connection of any sort to.my belief.

Thanks
Pavel