One document matched: draft-ietf-6man-impatient-nud-02.xml

<?xml version="1.0"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd" [
    <!ENTITY rfc0826 PUBLIC '' 
      'http://xml.resource.org/public/rfc/bibxml/reference.RFC.0826.xml'>
    <!ENTITY rfc2119 PUBLIC '' 
      'http://xml.resource.org/public/rfc/bibxml/reference.RFC.2119.xml'>
    <!ENTITY rfc3971 PUBLIC '' 
      'http://xml.resource.org/public/rfc/bibxml/reference.RFC.3971.xml'>
    <!ENTITY rfc4861 PUBLIC '' 
      'http://xml.resource.org/public/rfc/bibxml/reference.RFC.4861.xml'>
    <!ENTITY rfc6583 PUBLIC '' 
      'http://xml.resource.org/public/rfc/bibxml/reference.RFC.6583.xml'>
]>

<?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?>
<?rfc toc="yes"?>
<?rfc tocdepth="4"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc strict="no"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<?xml version="1.0"?>

<rfc ipr="trust200902" docName="draft-ietf-6man-impatient-nud-02.txt" updates="4861">
	
<front>
<title abbrev="NUD is too impatient">
Neighbor Unreachability Detection is too impatient
</title>

<author initials="E" surname="Nordmark" fullname="Erik Nordmark">
  <organization>Cisco Systems, Inc.</organization>
  <address>
    <postal>
    <street>510 McCarthy Blvd.</street>
    <city>Milpitas, CA, 95035</city>
    <country>USA</country>
    </postal>
    <phone>+1 408 527 6625</phone>
    <email>nordmark@cisco.com</email>
  </address>
</author>

<author initials="I" surname="Gashinsky" fullname="Igor Gashinsky">
  <organization>Yahoo!</organization>
  <address>
    <postal>
      <street>45 W 18th St</street>
      <city>New York, NY</city>
      <country>USA</country>
    </postal>
    <email>igor@yahoo-inc.com</email>
  </address>
</author>


<date month="Jul" year="2012"/>
<workgroup>6MAN WG</workgroup>
<keyword>6MAN</keyword>

<abstract>
<t>IPv6 Neighbor Discovery includes Neighbor Unreachability Detection. That 
function is very useful when a host has an alternative, for instance multiple
default routers, since it allows the host to switch to the alternative in
short time. This time is 3 seconds after the node starts probing by default. However,
if there are no alternatives, this is far too impatient. This document
specifies relaxed rules for Neighbor Discovery retransmissions that allows an
implementation to choose different timeout behavior based on whether or not there are alternatives.</t>
</abstract>
</front>

<middle>

<section title="Introduction">
<t>
IPv6 Neighbor Discovery <xref target="RFC4861"/> includes Neighbor 
Unreachability Detection (NUD), which detects when a neighbor is no longer 
reachable.The timeouts specified are very short (by default three transmissions 
spaced one second apart). That can be appropriate when there are alternative 
paths over which the packets can be sent. For example, if a host has multiple 
default routers in its Default Router List, or if the host has a 
Neighbor Cache Entry (NCE) created by a Redirect 
message. The effect of NUD reporting a failure in those cases is that the host 
will try the alternative; the next router in the Default Router List, or 
discard the NCE which will also send using a different router.</t>

<t>For that reason the timeouts in <xref target="RFC4861"/> were chosen to be 
short; this ensures that
if a default router fails the host can use the next router in less than 45 
seconds (taking into account a default ReachableTime of 30 seconds and the 
time spent in the DELAY state).</t>

<t>However, when there is no alternative there are several benefits in making
NUD try probing for a longer time. One of those benefits is to be more
robust against transient failures, such as spanning tree reconvergence and other
layer 2 issues that can take many seconds to resolve. Marking the NCE as 
unreachable in that case causes additional multicast on the network. Assuming
there are IP packets to send, the lack of an NCE will result in multicast 
Neighbor Solicitations every second instead of the unicast Neighbor 
Solicitations that NUD sends.</t>

<t>As a result IPv6 Neighbor Discovery is operationally more brittle than IPv4 ARP. For IPv4 there
is no mandatory time limit on the retransmission behavior for ARP 
<xref target="RFC0826"/> which allows implementors to pick more robust 
schemes.</t>

<t>The following constant values in <xref target="RFC4861"/> seem to have
been made part of IPv6 conformance testing: MAX_MULTICAST_SOLICIT, 
MAX_UNICAST_SOLICIT, and
RETRANS_TIMER. While such strict conformance testing seems consistent
with <xref target="RFC4861"/>, it means that we need to update the standard
if we want to allow IPv6 Neighbor Discovery to be as robust as
ARP.</t>

<t>This document updates RFC 4861 to relax the retransmission rules.</t>

<t>Additional motivations for making IPv6 Neighbor Discovery more robust
in the face of degenerate conditions are covered in
<xref target="RFC6583"/>.</t>

</section>

<section title="Definition Of Terms">

<t>
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL
NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and
"OPTIONAL" in this document are to be interpreted as described
in <xref target="RFC2119"/>.
</t>
</section>

<section title="Protocol Updates">

<t>Giving up after three packets spaced one second apart is only REQUIRED when
there is an alternative, such as an additional default route or a redirect.</t>

<t>If implementations transmit more than MAX_*CAST_SOLICIT packets it SHOULD use
(binary) exponential backoff of the retransmit timer. This is to avoid any
significant load due to a steady background level of retransmissions from 
implementations that try for a long time.</t>

<t>Even if there is no alternative, the protocol needs to be able to
handle the case when the link-layer address of the destination has changed by 
switching to multicast Neighbor Solicitations at some point in time.</t>

<t>In order to capture all the cases above this document introduces a new
UNREACHABLE state in the conceptual model described in <xref target="RFC4861"/>.
A NCE in the UNREACHABLE state retains
the link-layer address, and IPv6 packets continue to be sent to that link-layer
address. But in the UNREACHABLE state the NUD Neighbor Solicitations are
multicast, using a timeout that follows a (binary) exponential backoff.</t>

<t>In the places where RFC4861 says to to discard/delete the NCE after
N probes (Section 7.3, 7.3.3 and Appendix C) we will instead transition to
the UNREACHABLE state.</t>

<t>If the Neighbor Cache Entry was created by a redirect, a node MAY delete the
NCE instead of changing its state to UNREACHABLE. In any case, the node SHOULD
NOT use an NCE created by a Redirect to send packets if that NCE is in 
unreachable state. Packets should be sent following the next-hop selection
algorithm in section 5.2 in <xref target="RFC4861"/>
which disregards NCEs that are not reachable.</t>

<t>The default router selection in section 6.3.6 says to prefer default routers
that are "known to be reachable". For the purposes of that section, if the NCE
for the router is in UNREACHABLE state, it is not known to be reachable. 
Thus the particular text in section 6.3.6 which says "in any
state other than INCOMPLETE" needs to be extended to say "in any
state other than INCOMPLETE or UNREACHABLE".</t>

<t>Apart from the use of multicast NS instead of unicast NS, and the (binary)
exponential backoff of the timer, the UNREACHABLE state works the same
as the current PROBE state.</t>

<t>A node MAY garbage collect a Neighbor Cache Entry as any time as specified
in RFC 4861. This does not change with the introduction of the UNREACHABLE
state in the conceptual model.</t>

<t>The UNREACHABLE state is conceptual and not a required part of this 
specification. A node merely needs to satisfy the externally observable 
behavior of this specification.</t>

<t>There is a non-obvious extension to the state machine description 
in Appendix C in RFC 4861 in the case for "NA, Solicited=1, Override=0. Different link-layer
address than cached". There we need to add "UNREACHABLE" to the current list
of "STALE, PROBE, Or DELAY". That is, the NCE would be unchanged.
 Note that there is no corresponding change necessary to 
the text in section 7.2.5 since it is phrased using "Otherwise" instead of explicitly listing the three states.</t>

<t>The other state transitions described in Appendix C handle the introduction
of the UNREACHABLE state without any change, since they are described using
"not INCOMPLETE".
</t>

<t>There is also the more obvious change already described above. RFC 4861 
has this:</t>
<figure>
 <artwork><![CDATA[
PROBE           Retransmit timeout,     Discard entry         -
                N or more
                retransmissions.
]]></artwork>
</figure>

<t>That needs to be replaced by:</t>
<figure>
 <artwork><![CDATA[
PROBE           Retransmit timeout,     Double timeout    UNREACHABLE
                N or more               Send multicast NS
                retransmissions.

UNREACHABLE     Retransmit timeout      Double timeout    UNREACHABLE
                                        Send multicast NS
]]></artwork>
</figure>

<t>The binary exponential backoff SHOULD be clamped at some reasonable
maximum retransmit timeout, such as 60 seconds. If there is no IPv6
packets sent using the UNREACHABLE NCE, then it makes sense to stop the
retransmits of the multicast NS until either the NCE is garbage collected
or there are IPv6 packets sent using the NCE. 
The multicast NS and associated binary exponential backoff
can be applied on the condition of the continued use of the NCE to send IPv6
packets to the recorded link-layer address.</t>

<t>A node MAY unicast the first few Neighbor Solicitation messages
while in UNREACHABLE state, but it MUST switch to multicast Neighbor 
Solicitations. Otherwise it would not detect a link-layer address change
for the target.</t>

</section>

<section title="Example Algorithm">

<t>This section is NOT normative, but specifies a simple implementation which
conforms with this document. The implementation is described using operator
configurable values that allows it to be configured in a way to be
compatible with the retransmission behavior in <xref target="RFC4861"/>.
The operator can configure the values for MAX_*CAST_SOLICIT, RETRANS_TIMER,
and the new BACKOFF_MULTIPLE and MARK_UNREACHABLE. This allows the
implementation to be as simple as:
</t>

<t>
next_retrans = ($BACKOFF_MULTIPLE^$solicit_attempt_num)*$RetransTimer +
jittered value.
</t>
<t>
After MARK_UNREACHABLE retransmissions the implementation would mark the 
NCE UNREACHABLE and switch to multicast NUD probes.
</t>

<t>
The recommended behavior is to have 5 attempts, with timing spacing of 0
(initial request), 1 second later, 3 seconds later, then 9, and then 27, and
switch to UNREACHABLE after 3 transmissions, which represents:
<list>
    <t>MAX_UNICAST_SOLICIT=5</t>

    <t>RETRANS_TIMER=1 (default)</t>

    <t>BACKOFF_MULTIPLE=3</t>

    <t>MARK_UNREACHABLE=3</t>
</list>
After 3 retransmissions the implementation would mark the NCE UNREACHABLE
and switch to multicast NUD probes. Thus we enter UNREACHABLE, and try any 
available alternative, after 4 seconds compared to the current 2 seconds. That
additional delay is small compared to the default 30 seconds ReachableTime.
</t>

<t>
If BACKOFF_MULTIPLE=1, MARK_UNREACHABLE=3 and
MAX_UNICAST_SOLICIT=3, you would get the same behavior as in
<xref target="RFC4861"/>.
</t>

<t>
An Implementation following this algorithm would, if the request was
not answered at first due for example to a transitory condition,
retry immediately, and then back off for progressively longer
periods.  This would allow for a reasonably fast resolution time when
the transitory condition clears. 
</t>

<t>
Note that RetransTimer and ReachableTime are by default set from
the protocol constants RETRANS_TIMER and REACHABLE_TIME, but are
overridden by values advertised in Router Advertisements as specified
in <xref target="RFC4861"/>. That remains the case even with the 
protocol updates specified in this document.
The key values that the operator would configure are BACKOFF_MULTIPLE,
MAX_UNICAST_SOLICIT and MAX_MULTICAST_SOLICIT.
</t>

<t>It would be useful to have a maximum value for 
($BACKOFF_MULTIPLE^$solicit_attempt_num)*$RetransTimer
so that the retransmissions are not too far apart. A value 60 seconds is
consistent with DHCP.</t>

</section>

<section title="Acknowledgements">

<t>The comments from Thomas Narten, Philip Homburg, and Joel Jaeggli
have helped improve this draft.</t>

</section>

<section title="Security Considerations">

<t>Relaxing the retransmission behavior for NUD is belived to have no impact on
security.  In particular, it doesn't impact the application Secure Neighbor
Discovery <xref target="RFC3971"/>.</t>

</section>

<section title="IANA Considerations">

<t>This are no IANA considerations for this document.</t>

</section>

</middle>

<back>
<references title="Normative References">
&rfc2119;
&rfc3971;
&rfc4861;
</references>

<references title="Informative References">
&rfc0826;
&rfc6583;
</references>
</back>
</rfc>