<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p><br>
    </p>
    <div class="moz-cite-prefix">On 15.08.2019 00:50, teor wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:850F7D16-A937-4747-9D27-905850C9034F@riseup.net">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div><span style="background-color: rgba(255, 255, 255, 0);">Hi,</span></div>
      </div>
      <div dir="ltr"><br>
        On 14 Aug 2019, at 03:42, NOC <<a
          href="mailto:tor@afo-tm.org" moz-do-not-send="true">tor@afo-tm.org</a>>
        wrote:<br>
        <br>
      </div>
      <blockquote type="cite">
        <div dir="ltr">
          <div class="moz-cite-prefix">On 12.08.2019 23:39, teor wrote:</div>
          <blockquote type="cite"
            cite="mid:06830E45-8E86-476C-A08B-262FD54DEAA4@riseup.net">
            <div dir="ltr"><br>
              On 13 Aug 2019, at 05:08, Roman Mamedov <<a
                href="mailto:rm@romanrm.net" moz-do-not-send="true">rm@romanrm.net</a>>
              wrote:<br>
              <br>
            </div>
            <blockquote type="cite">
              <div dir="ltr"><span>On Mon, 12 Aug 2019 00:46:50 +0000</span><br>
                <span>Christopher Sheats <<a
                    href="mailto:yawnbox@emeraldonion.org"
                    moz-do-not-send="true">yawnbox@emeraldonion.org</a>>
                  wrote:</span><br>
                <span></span><br>
                <blockquote type="cite"><span>Tor Project, please
                    increase your #IPv6 awareness/outreach similar to
                    how</span><br>
                </blockquote>
                <blockquote type="cite"><span>ARIN and the other RIRs
                    try very hard to do.</span><br>
                </blockquote>
                <span></span><br>
                <span>Before outreach Tor would need some actual IPv6
                  support, as in using it for</span><br>
                <span>the actual traffic of relay-to-relay
                  communication. I tried running a few</span><br>
                <span>relays with very fast IPv6 and slow IPv4 (due to a
                  common NAT frontend which</span><br>
                <span>was the bottleneck), but it was a complete
                  nonstarter.</span></div>
            </blockquote>
            <div><br>
            </div>
            <div>Tor relays currently don't connect over IPv6. When 10%
              of the network</div>
            <div>supported IPv6, there wasn't much point, because
              putting a very small</div>
            <div>number of paths over IPv6 has privacy risks. So we
              focused on client, guard,</div>
            <div>and exit IPv6 support.</div>
            <div><br>
            </div>
            <div>But <span style="background-color: rgba(255, 255, 255,
                0);">currently, about 30% </span><span
                style="background-color: rgba(255, 255, 255, 0);">of the
                consensus </span><span style="background-color:
                rgba(255, 255, 255, 0);">weight supports IPv6. So we</span></div>
            <div><span style="background-color: rgba(255, 255, 255, 0);">are
                working on a grant for IPv6 support (see below).</span></div>
            <div><br>
            </div>
            <div>We won't be able to prefer IPv6 until 50-67% of relays
              support IPv6, for</div>
            <div>load-balancing and privacy reasons.  But we plan on
              using the</div>
            <div>"Happy Eyeballs" (RFC 8305) algorithm on dual-stack
              relays. So</div>
            <div>sufficiently slow IPv4 will cause relays to connect
              over IPv6. (And we can</div>
            <div>tune the load-balancing using the IPv4 to IPv6 delay.)</div>
          </blockquote>
          <p>I still would say that these stats are deeply flawed.
            Looking at the Autonomous Systems where the relays are
            located from the top100, 99 of them do support IPv6
            (85,7625% consensus weight), the only one which doesn't
            support is AS4224 but since they manage their AS themselves
            they would only need to ask their LIR and would get IPv6.</p>
        </div>
      </blockquote>
      <div>The top 100 relays are only 13-18% of the total advertised
        bandwidth:</div>
      <div><a
href="https://metrics.torproject.org/advbwdist-relay.html?start=2019-05-16&end=2019-08-14&n=1&n=100"
          moz-do-not-send="true">https://metrics.torproject.org/advbwdist-relay.html?start=2019-05-16&end=2019-08-14&n=1&n=100</a></div>
      <div><a href="https://metrics.torproject.org/bandwidth.html"
          moz-do-not-send="true">https://metrics.torproject.org/bandwidth.html</a></div>
    </blockquote>
    <p>I never wrote about the top100 relays, relays don't matter, they
      come and go. It is important who does host them, that is why i
      looked at the AS, because the providers won't stop offer IPv6 if
      they have deployed it once. And that is why i think the complete
      roadmap is not useful at all and will delay everything just
      unnecessary.<br>
    </p>
    <blockquote type="cite"
      cite="mid:850F7D16-A937-4747-9D27-905850C9034F@riseup.net">
    </blockquote>
  </body>
</html>