I switched to Railway, at least for now. It was a bit more work than with Galaxy to get up and running, having to set up MongoDB and having to figure out how to connect it to the Node instance.
The reason for the switch is that Galaxy currently has intermittent network failures across regions reported by users, and verified by myself using VPN. After switching to Railway, there are no network failures, and the same users who had issues can now access the app.
The Railway management system (apart from having to do some custom work for Meteor setup because it does not natively know about Meteor, but it knows about Node.js and MongoDB separately, requiring you to deploy the Meteor Node.js code as one service and then connect a separate MongoDB to it as a separate service), is that all the services you deploy for a project are on a visual canvas with floating boxes for each service, and it shows lines connecting the boxes to help you visualize exactly how your internal network is set up.
It has a variable system to allow any services to connect to other services, where variables from one service (such as MONGO_URL for the MongoDB service) can be used by another service (f.e. the Node.js service), and this is what causes a line to connect the Node.js service to the MongoDB service in the visual canvas. For example my Node.js service has a variable MONGO_URL that has its value set to ${{MyMongoDBService.MONGO_URL}} which will automatically fill in a MONGO_URL env var for the Node.js service when it is launched.
This is very interesting because I can configure, or replace the MongoDB with a new one, and the URL will magically stay updated behind the scenes (no need to copy/paste mongodb://user:pass@domain.example.com URLs around.
There are also global variables. Why this is powerful: you can define vars like PROD_URL or DEV_URL, and then you can make multiple app instances in multiple regions, each as represented as visual nodes on your visual canvas, and you can easily set ROOT_URL to ${{PROD_URL}} or ${{DEV_URL}}, and use it as a template. Any time you replicate a node into a new region, or in the same region, it can use the correct global URL.
Then, you can change the shared variables globally, restart the containers, and they all use the new URL! So this makes management of config really easy.
All of the Railway configs can be exported and saved into a JSON file in your repo, giving you more control (f.e. deploy a new instance from git, and it will use that config file rather than via the Railway UI).