Re: [Gzz] hemppah's research problems document

[Tuomas Lukka]

> >>>
> >>>Some systems, such as DHTs cannot let other people use this "mirror" 
> >>>because
> >>>the files are not in the location the DHT would place them in.
> >>>
> >>>So if we think about a 5GB / 100MB split between my mirrored data / DHT 
> >>>area (which is quite reasonable), most of the potential capacity in the 
> >>>network
> >>>is not getting used!
> >>
> >>I don't understand at all. As I have been so adamant about in my
> >>statements to hemppah, the DHT is for *locating* the data in the
> >>network. 
> >
> >
> >No, that raises entirely different set of issues.
> 
> An entirely different set of issues than what?
> 
> >>Your machine would, when going online, place into the DHT
> >>mappings (blockid -> your-ip-address), so if any computer tries to
> >>download the RFCs, they would in the DHT get your machine's address, and
> >>then contact your machine to download from there. 
> >
> >The problem is that this puts in more steps to the algorithm and makes it 
> >MUCH more vulnerable to attacks: simply placing bogus data will be a pretty
> >good attack.
> 
> More vulnerable to attacks than what?
> 
> Are you *seriously* suggesting that the DHT should store the actual blocks?

What I'm seriously saying is that 

1) DHT storing the actual blocks is
        + pretty well understood
        + quite efficient, robust and not too attackable
        - not good for us, because data is where the network wants it,
          not where users want it to be
2) DHT storing IPs of nodes is
        + in some ways what we'd like, because data is where users
          want it
        - not well understood and researched
        - efficiency unknown, if nodes blink in and out:
                in a DHT, the nodes nearby would notice that some node
                is down and remove it from the routing tables. In your
                suggestion ("benjanet" ;), the actual IP would be what
                everyone gets and tries to access.
        - unknown if we can ensure that data is reasonably replicated
        - quite easy to attack in various ways

I'm saying that the assumption, when talking about DHTs is 1), since
that's the alternative that's known.

2) is not at all known, and has potentially huge drawbacks.

        Tuomas