RE: Opportunistic Tunneling

["Jeroen Massar" <jeroen@unfix.org>]

-----BEGIN PGP SIGNED MESSAGE-----

Pekka Savola wrote:

<SNIP>

A small non-techy, economic story...

> such as the economic (non-incentive) for deploying such an anonymous 
> tunnel service, which would probably lead to a lot of trouble in the 
> long run (abuse reports from your netblock, increased traffic ~= 
> bigger payments to your transit operators, etc.).

Seeing Tunnel Brokers like Freenet6, Hurricane, BTExact and SixXS being
around for more than 3 years now does seem to answer the question if
it is still feasible to provide such a service. The ISP's involved in
these projects who are donating the bandwidth, resources and probably
the most costly of them all, manpower, apparently have a benefit or
don't see it as a big economic burden to run these services.
I don't know how the other TB's are run, but SixXS is run purely on
good will, the bandwidth and hardware are sponsored by the ISP's who
run the POP. Currently it is doing about 10mbit/s in total across the
POP's we run and most traffic stays local (non-transit) as we have
multiple POP's in various countries and users get directed to the
network-wise closest POP which is better for latency and costs.

The ideal situation, like 6to4 relay's and webcaches etc. would of
course be when every ISP would run an instance of a TB or 6to4 relay
etc thus spreading the economic burden. Currently we try to keep
users as local as possible, but countries like Poland and Italy have
apparently a big userbase that want IPv6 but no (good) TB/6to4's
there, as they come to us.

Simple calculation of costs for the ISP having users wanting IPv6
and not having any IPv6 TB/6to4-relay: traffic will flow to other
ISP's, these ISP's in turn will have the cost of inbound traffic
and the cost of the TB/relay to the remote IPv6 host. When the
TB/relay is local the ISP connecting the enduser won't have much
costs as it is usually transit, when the traffic becomes transit
and larger volumes the ISP will notice. Either way the ISP's who
do not offer a TB/6to4 relay instance cheaply lift along with
the ones that do.

Thus it would be 'better' for the deployment when multiple
instances of TB's or 6to4 relays would be deployed thus spreading
the costs over these multiple ISP's. Currently there is still not
much traffic and the traffic that is there is mostly local thus
I don't think there is a big incensitive to do this for most ISP's
and that really counts for the ISP's who don't want to take some
time for even looking at IPv6 at all. But when the traffic rates
will rise and the bigger sites start being available over IPv6
it will become costly to run TB's and 6to4's for the ISP's that
are doing it for free now.

Small tech note: we need to have this done seemlessly and easy
for the enduser.

As for the abuse reports: many TB's prohibit access to IRC servers
by blocking ports 6600-7100 or something, the others handle a
'give us valid address and contacting data' and just instantly
shutdown abusers on the first warning, doing the latter does make
it more difficult to get a service but does keep the service a
quality one and keeps abuse low and almost near nill.

Greets,
 Jeroen

-----BEGIN PGP SIGNATURE-----
Version: Unfix PGP for Outlook Alpha 13 Int.
Comment: Jeroen Massar / http://unfix.org/~jeroen

iQA/AwUBQDqdSymqKFIzPnwjEQKe/wCfXdIqXmCJjhrsJZA8rCzDop3DRIgAnRGx
WEPtXedbOme/lcS8faYZCVmS
=IzBF
-----END PGP SIGNATURE-----