Can ray-core work over decentralized p2p network created by libp2p?
Which language is more adequate for that?
Please code with examples without Ray[AIR]
I have my own p2p decentralized(gossip) network created with libp2p and must to use RAY-core
to work simple AI task using that existing p2p network.
In conclusion must to use Ray together with libp2p.
Which language need to use and how show the simple code?
Can you describe what functionality libp2p provides? for example, is there the notion of nodes opening ports for listening and connecting?
Probably this is doable but I doubt it is easy or plug-and-play.
Yes…I will use it to do peer to peer network based on go.
You can see about it : Implementations - libp2p
I tried that application : GitHub - mudler/edgevpn: The immutable, decentralized, statically built p2p VPN without any central server and automatic discovery! Create decentralized introspectable tunnels over p2p with shared tokens as demo and work ok
So it’s simple to create the p2p network in various languages.
After that need to use Ray to work over this network…
Ray need to attach the p2p network and then need to work in paradigm Mater-workers for AI task.
With other words need to use Ray for distributed AI task but using libp2p’s network.
Okay. Probably you can start here for the ports that Ray needs to operate Configuring Ray — Ray 2.2.0.
The Ray have one implementation in GO lang?
I not want or intend to use one cluster from ray or other resources from ray.
I want to use my node’s resources from my private(isolate) peer2peer(p2p) network
and in this case how can do that from ray?
So need to use ray but on my private libp2p network… that ideea!
Can be done that from ray and in which langauage go,python or java?
So need to create one distributed task(like AI or any distributed calculation) but using resourses from my private p2p network(already created with libp2p)
Some code like:
import ray
ray.init(address=“:10001”, _system_config={“env”: {“TOKEN”: “”}})
@ray.remote
def add(x, y):
return x + y
result_ids =
for i in range(10):
for j in range(10):
result_ids.append(add.remote(i, j))
result = sum(ray.get(result_ids))
print(result)
can be used in this form from python? or … ray need/must to have access to my private routing table(because my routing table was stored into swarm’s blockchain and for local nodes I used mDNS and for internet nodes I used Kademlia DHT protocol…in plus the my private network was created using libp2p from GO/golang)?
did you run ray on p2p networks successfully? i am also interested in this idea…