protolambda
Question regarding DHTs
Hey proto,
pretty new to p2p protocols, so trying to wrap my head around some of the things you mention in the README, such as DHTs are slow, right? and Number of hops will be low if some type of DHT search even works only a little bit.. I assumed the first comment was rhetorical and indicating that you had a way to make them fast, but the second comment seems to contradict that viewpoint.
I've also found this comment that mentions
with 64 subnets in discv5, it takes ~30 secs to discover around 2-3 peers in the required subnet.
With a 32x increase in subnets, I'm not sure if it would work as well.
Are we looking into any alternative topic based discovery mechanisms for subnet discovery?
which make DHT sound like the most useless thing in the world.
So the solution here is to create an overlay DHT, meaning a subset of discv5 nodes all of which are participating in the DAS sampling protocol? What's the point of using discv5 in the first place if we don't communicate with most of the nodes in there? Is the point that creating an overaly on top of a DHT easier than just defining a new DHT from scratch?
