6 ms·

How does this compare with cjdns? Also have you looked into eventual consistency?

api 12y ago
They're different things architecturally. cjdns is cool but as far as I know could not scale really really huge or survive, say, a Sybil type attack by someone with NSA or top-tier black hat hacker group expertise. I'm not sure my system could survive that either, but the problem is I'm not convinced a centerless mesh can be made that robust. I know some community mesh nets have used it with some success. Just because something has limits doesn't mean it can't be very useful for lots of applications.
- CMCDragonkai 12y ago
  You mentioned the problems with CAP theorem. Did you investigate methods to use eventual consistency in AP system?
  - api 12y ago
    With eventual consistency "eventual" becomes a long time with a big network. Either that or the bandwidth requirements become massive. It would be an example of choosing security and decentralization at the expense of efficiency.
    - CMCDragonkai 12y ago
      Would it be possible to make zerotier tunable so in some networks you care about efficiency, in others you could make it eventually consistent. I know there are lots of different methods for eventual consistency that could be used and compared for performance.
      - api 12y ago
        Tunable means there's now something else people have to understand, which means nobody will use it.
        CMCDragonkai 12y ago
        I'm sure the zero-conf can be default. But for some custom enterprisy work or projects, having something tunable for eventual consistency is something I would use.
        api 12y ago
        Multicast presently does have some tunable parameters, but so far they all have acceptable defaults and are hidden. I will probably expose them under an advanced tab in the future. So I didn't mean to be totally dismissive... My point was just that every tunable parameter a user has to interact with loses you an exponential number of users.
- CMCDragonkai 12y ago
  What if the entire network was a VPN where you the operator know all the nodes that are part of it. Would it still be vulnerable to the sybil attack?
  - walterbell 12y ago
    Related: Tinc open-source VPN with some P2P properties, http://blogs.operationaldynamics.com/andrew/software/research/using-tinc-vpn http://blogs.operationaldynamics.com/andrew/software/researc...

api 12y ago

They're different things architecturally. cjdns is cool but as far as I know could not scale really really huge or survive, say, a Sybil type attack by someone with NSA or top-tier black hat hacker group expertise. I'm not sure my system could survive that either, but the problem is I'm not convinced a centerless mesh can be made that robust. I know some community mesh nets have used it with some success. Just because something has limits doesn't mean it can't be very useful for lots of applications.

CMCDragonkai 12y ago
You mentioned the problems with CAP theorem. Did you investigate methods to use eventual consistency in AP system?
- api 12y ago
  With eventual consistency "eventual" becomes a long time with a big network. Either that or the bandwidth requirements become massive. It would be an example of choosing security and decentralization at the expense of efficiency.
  - CMCDragonkai 12y ago
    Would it be possible to make zerotier tunable so in some networks you care about efficiency, in others you could make it eventually consistent. I know there are lots of different methods for eventual consistency that could be used and compared for performance.
    - api 12y ago
      Tunable means there's now something else people have to understand, which means nobody will use it.
      - CMCDragonkai 12y ago
        I'm sure the zero-conf can be default. But for some custom enterprisy work or projects, having something tunable for eventual consistency is something I would use.
        api 12y ago
        Multicast presently does have some tunable parameters, but so far they all have acceptable defaults and are hidden. I will probably expose them under an advanced tab in the future. So I didn't mean to be totally dismissive... My point was just that every tunable parameter a user has to interact with loses you an exponential number of users.
CMCDragonkai 12y ago
What if the entire network was a VPN where you the operator know all the nodes that are part of it. Would it still be vulnerable to the sybil attack?
- walterbell 12y ago
  Related: Tinc open-source VPN with some P2P properties, http://blogs.operationaldynamics.com/andrew/software/research/using-tinc-vpn http://blogs.operationaldynamics.com/andrew/software/researc...