The description doesn't say how items are transferred from platform to platform - teleportation, shared container, rockets, etc. I'm just curious how fast items transfer from point to point.

Hello, they use cargo pods. The same cargo pods that platforms use to drop items to planet. The rate limit is determined by how busy a hatch is. Basically, it's pretty high. Whenever a hatch is considered "not busy", it can send a pod.

Ah. Thank you kindly for the explanation. That's pretty much the best way of handling it, I think. I'll have to look up what constitutes a "busy" hatch and how to calculate throughput, but that's on me to figure out :)

My understanding is that a hatch is busy when it's reserved by an incoming pod, or it has reached the maximum amount of pods that can be present inside it at an instant.
https://lua-api.factorio.com/2.0.70/classes/LuaCargoHatch.html#busy

The doc doesn't say anything about the definition of busy.

Also, for IPL, there is a mod setting to increase the amount of requests that can be processed per tick. Changing it may increase the throughput in the case where you aready have many requests.

What's your endgame ?

What's your endgame ?

Not sure what you mean by "end game" here, but what I'm trying to achieve is... Sort of an orbital storage platform sitting over Nauvis, collecting resources which are slow to send up (like nuclear ammo), from which other platforms can pull when they swing by to restock. This way, ground silos can send resources up into space constantly, rather than only when a platform is parked over the planet. Hoping to speed up sending items to other planets, basically :)

And thank you for the explanation. I was curious if adding more Cargo Bays would speed up throughput, as each one comes with its own hatch - to make a long story short.

Yep that's what I meant by endgame, "end goal", or "true question" :D.

Yes more hatches = more throughput, so adding cargo bays will help. Althought, the mod tries to fill every slot of every pod it sends, (a pod has 10 inventory slots) which means a lot of ressource for a single pod, it will also try to send as many pods as possible whenever processing a request. Overall it's pretty fast. I haven't measured the throughput exactly, but I would be curious to see anyone find it "too slow".