Some of these operations can be cacheable which is ideal for scalability, but even the ones that cannot tend to be far cheaper than running VMs which incur by the minute CPU costs.
Our intention is to be fully transparent on the how & why, so I’m happy to answer further questions on all this! I also think it’d be good for us to include something more detailed like your explanation in our docs so that other folks don’t feel similarly confused when reading them- curious if there’s particular points that jump out to you that I can share with our team to include!
