Terminology question: Peer Store versus DHT

[ross-spencer]

Quite new to libp2p and have been doing a lot of reading and trying to understand the examples but it's slow going.

I have a closed network that will be max 100 nodes (known IP and ports -- their identity might be changed manually and updated dynamically (and cryptographically validated) outside of the node's init functions.

I believe, at a minimum,

• each node must have a known and persistent identity.
• each identity must be loaded into a DHT.
• the DHT must be provided to a discovery function.

This will then allow me to connect each node to each other node and data can be sent to all.

i.e. that each node becomes a bootstrap node, and also a "participating" node? (again terminology might not be right)

Does the terminology here sound correct? And is is a DHT the same as a Peer Store? Do I load identities into a DHT, or a Peer Store?

One similar demo app in the code-base is the routed-echo, is that a good point of reference for this type of app?

[MarcoPolo]

The peer store is a local database for your local view of peers on the network. The DHT is a distributed database. If you know all the peers on the network, then you should load those identities into the peer store.

[ross-spencer]

Is it possible to initialize an empty peer store and add to it? Do you know how that would then be supplied to a node when starting it for the first time?

[MarcoPolo]

Once you have your [Host](https://pkg.go.dev/github.com/libp2p/[email protected]/core/host#Host) you can call `.Peerstore()`. Refer to the go docs from there.;

[ross-spencer]

What am I missing? Is there a .Add() function for the peers?

Also, clearly "following the docs" isn't the most helpful if I am here in the Q&A... do you know of any demo apps that achieve this?

[MarcoPolo]

apologies for the formatting. Apparently replying via email has some bugs on the GitHub side. Hopefully the fixed formatting explains it better.