I want to replace the core of my project with Iroh, but I have some concerns.

[ISensuiI]

First of all, I want to state that as a user, I am very grateful to the developers and contributors of the two projects, libp2p and Iroh. My discussion below does not involve belittling a certain project. I am just exploring a solution that is more suitable for me.

Since my current project is based on rust+libp2p, the code volume is about 30,000+, and it will be around 40,000-50,000 in the future (60% completed). So this switch is a very large project. And the current project is an experimental project. Later this year, we will use Iroh to build another more complex project. So there are a lot of problems.

Reasons for migration:

1. My previous project was built with libp2p, and its main function was multiplayer remote communication (games). The most important issue here is hole punching, because if the hole punching success rate is low, it will first lead to high latency and heavy load on the relay server.

2. Mobile terminal binding support I noticed that Iroh has better support for various language bindings

I have learned about nat, hole punching, several types of nat, and the principles of each type of hole punching. And I have read Iroh All relevant documents, but there are still some questions.

Question 1:
Regarding the success rate of hole punching, I have recently learned about the Iroh project in many aspects. Through grok3 analysis, it tells me that the success rate of Iroh hole punching is usually 90% (I remember that Iroh has a page showing the success rate of hole punching, but I can't find it), while libp2p is 70-80% (Grok DeepSeek and other AI comprehensive analysis). I don't have a standard data reference for libp2p (I have asked multiple AIs to give me my comprehensive ratio). I asked a question a month ago and learned that the core developers of Iroh are all deep users of libp2p. So I want to know whether Iroh's hole punching success rate is completely better than libp2p in several network types?

Question 2:
In the libp2p project, I mainly use the Requestresponse protocol for relay servers and multiple nodes to communicate. I understand that there is no such protocol in Iroh. But I know there is iroh-docs. Can I use iroh docs instead of RR protocol? Used for various node status storage and information synchronization between nodes (I know there is gossipsub. If docs cannot be used for real-time synchronization of node status, I may use gossipsub)

Question 3:
My current project is divided into relay center server A (single node), relay edge server BCDE.. (multiple nodes), A is used to route and distribute user requests to other relay nodes. Use rendezvous protocol for processing. Iroh I noticed that there is no relevant protocol. This needs to be built by yourself, right? Is there any problem with this architecture Iroh?

Question 4:
I want to implement Full Mesh Network between users, using quic sub-stream multiplexing? I didn't see any relevant description on the official website. I know this question is stupid, but I still want to confirm that Iroh can implement Full Mesh Network without any problem, right?

Question 5:
Iroh hole punching requires the use of DERP related technology, which I don't know much about https://www.iroh.computer/blog/what-is-derp I just read some introductions here, I want to know, Normal P2P connection
Client A
(Relay + Signaling) B
Client C
These three ends can complete the connection. I have asked AI before and read the manual. Iroh has a default third-party derp server, right? The question is, assuming that our region (china) is sensitive to the network (many regions cannot be accessed), can I build a derp node myself?
Sorry, this may be a stupid question. In fact, I have not used Iroh yet. I can't tell the difference between the relay and the Derp service. Are they together, that is, our binary server file.
When I use libp2p, it only has one Relay.
More importantly, if I build it myself (not using the default Iroh Derp node), is the success rate of hole punching consistent with our Iroh default?

Question 6:
I have observed that Iroh updates very quickly, and most of them are destructive updates. I want to know how long this situation will last, 1 year or 2 years, Probably, I'm not putting pressure on the development team, I just want to confirm so that I can adjust my development plan. Because I have two very important projects in the future that I want to use Iroh.

[matheus23]

Hey, sorry for not answering all of your questions, but specifically about some parts:

• About relays / DERP: ("DERP" is an old term for our relay servers) Read more about them this page: https://www.iroh.computer/docs/concepts/relay
  and about self-hosting in our FAQ: https://www.iroh.computer/docs/faq#what-if-number-0-stops-running-relay-servers
• About breaking changes to iroh: We're planning for a 1.0 in July. We expect that there will be protocol-breaking changes until then, but not after. See also our roadmap: https://www.iroh.computer/roadmap

[ISensuiI]

OK, thanks for your reply. Regarding the success rate of drilling holes, there is no relevant data, right? (iroh) In fact, this is a very important indicator for me.