[Freedombox-discuss] Freedombox threat model

[Ted Smith]

On Wed, 2011-06-29 at 13:46 -0400, Daniel Kahn Gillmor wrote:
> On 06/29/2011 01:35 PM, Ted Smith wrote:
> > Any time two programs running on FreedomBoxes are talking to each other,
> > there's very little reason not to use Tor. It only makes sense not to
> > use Tor if there's no way to anonymize the protocol, in cases like
> > BitTorrent, or if there's way too much bandwidth required for Tor to
> > provide, like video streaming.
> 
> Other reasons to not use tor for certain workloads/environments:
> 
>  * you're not on an IPv4 network (tor only supports IPv4 at the moment,
> iiuc)
> 
>  * your protocol uses UDP or something else other than TCP (tor only
> does TCP at the moment, iiuc)
> 

An AF-independence patch just landed in Tor's main repo, so IPv6 should
be here Real Soon Now. I don't think there's any plan on supporting
anything other than well-formed TCP streams.

>  * you have tight latency constraints (not just throughput).  For
> example, the subjective experience of telephone calls is severely
> degraded by even a 150ms lag, even if the throughput is minimal.  Tor's
> extra hops add to the latency of any circuit.
> 
> 	--dkg

Tor is pretty good with throughput, but does poorly on latency. I don't
think I communicated that well.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.alioth.debian.org/pipermail/freedombox-discuss/attachments/20110630/f01d9263/attachment.pgp>