I have been thinking something similiar to this to replace IRC servers. The current way of having lots of IRC networks is just annoying, everything would work so much more nicely if clients could just connect to "IRC" and talk to everyone.
I've written up some of my thoughts but haven't really gotten around to do any code, mostly because currently I'm not IRCing that much and have several other projects to do.
This is pretty much what I was thinking of implementing a while ago, but this would require pretty huge servers at the top and I think I've found a better way.
Here's a paper I found from decentralization mailing list, it's a hash table distributed to each node in the network. It's reliable, fast enough and scalable. I've thought a few additions to it which would allow storing the content inside the hash too and how to search for data. The original authors may have better ideas but at least I know it could be made to work:)
Yes, SILC is good:) I've been playing it with a few months now and making my own IRC client (irssi) support it.
For me the best thing about SILC is not just the security, but that you can still affect it's development and the protocol to create the best IRC replacement you can think of. And I do intend the SILC to replace IRC one day:)
I guess I'm too late and no-one sees this message anymore, but anyway..:)
Take a look at SILC, Secure Internet Live Conferencing. It's designed with better network structure, isn't a braindead protocol, and as the name says it's designed with security in mind. And to me the best thing about it is that it's new and not finished yet. I can suggest new features to it, I can fix broken things in it, I can try to make it the best chat protocol there is.
SILC is the most serious IRC replacement I've seen so far: there's working server and client code, there's documentation, even comments in the code and the specs are in RFC drafts.
The biggest reason for DOS attacks against IRC servers is (I'm pretty sure:) creating net splits and taking over channels with them. If we just design the protocol so that it is impossible to take over others' channels the network will be DOS safe (and it will be one happy chatting network;)
I've been thinking about this for a few weeks. And it really needs a lot more thinking:) But I'm pretty sure it's possible, and could be made fast enough. And I'm thinking it as of a replacement for IRC, with channels and all. Everything would be based on public key cryptography.
I've set up a page for this in http://irssi.org/dchat/, the current idea paper is a bit old however, the flood routing can't work with huge networks:) Dividing the network to smaller groups would be better.
Slashdot Mirror
I have been thinking something similiar to this to replace IRC servers. The current way of having lots of IRC networks is just annoying, everything would work so much more nicely if clients could just connect to "IRC" and talk to everyone.
I've written up some of my thoughts but haven't really gotten around to do any code, mostly because currently I'm not IRCing that much and have several other projects to do.
Here's a paper I found from decentralization mailing list, it's a hash table distributed to each node in the network. It's reliable, fast enough and scalable. I've thought a few additions to it which would allow storing the content inside the hash too and how to search for data. The original authors may have better ideas but at least I know it could be made to work :)
For me the best thing about SILC is not just the security, but that you can still affect it's development and the protocol to create the best IRC replacement you can think of. And I do intend the SILC to replace IRC one day :)
Take a look at SILC, Secure Internet Live Conferencing. It's designed with better network structure, isn't a braindead protocol, and as the name says it's designed with security in mind. And to me the best thing about it is that it's new and not finished yet. I can suggest new features to it, I can fix broken things in it, I can try to make it the best chat protocol there is.
SILC is the most serious IRC replacement I've seen so far: there's working server and client code, there's documentation, even comments in the code and the specs are in RFC drafts.
The biggest reason for DOS attacks against IRC servers is (I'm pretty sure :) creating net splits and taking over channels with them. If we just design the protocol so that it is impossible to take over others' channels the network will be DOS safe (and it will be one happy chatting network ;)
I've set up a page for this in http://irssi.org/dchat/, the current idea paper is a bit old however, the flood routing can't work with huge networks :) Dividing the network to smaller groups would be better.