IPv6 Flow Label Specification
Level 3 Communications, LLC
1025 Eldorado Blvd
Broomfield
CO
80021
USA
shane@level3.net
Department of Computer Science
University of Auckland
PB 92019
Auckland
1142
New Zealand
brian.e.carpenter@gmail.com
Huawei Technologies Co., Ltd
Huawei Building, No.3 Xinxi Rd.,
Shang-Di Information Industry Base, Hai-Dian District, Beijing
P.R. China
shengjiang@huawei.com
Nokia-Siemens Networks
TBD
TBD
Finland
jarno.rajahalme@nsn.com
Internet
6MAN
This document specifies the IPv6 Flow Label field and the minimum
requirements for IPv6 nodes labeling flows, IPv6 nodes
forwarding labeled packets, and flow state establishment methods.
Even when mentioned as examples of possible uses of the flow
labeling, more detailed requirements for specific use cases are out
of scope for this document.
The usage of the Flow Label field enables efficient IPv6 flow
classification based only on IPv6 main header fields in fixed
positions.
A flow is a sequence of packets sent from a particular source to a
particular unicast, anycast, or multicast destination that a node
desires to label as a flow. A flow could consist of all packets in a
specific transport connection or a media stream. However, a flow is
not necessarily 1:1 mapped to a transport connection.
Traditionally, flow classifiers have been based on the 5-tuple of the
source and destination addresses, ports, and the transport protocol
type. However, some of these fields may be unavailable due to either
fragmentation or encryption, or locating them past a chain of IPv6
extension headers may be inefficient. Additionally, if classifiers
depend only on IP layer headers, later introduction of alternative
transport layer protocols will be easier.
The usage of the 3-tuple of the Flow Label and the Source and
Destination Address fields enables efficient IPv6 flow
classification, where only IPv6 main header fields in fixed positions
are used.
The flow label could be used in both stateless and stateful scenarios.
A stateless scenario is one where a node that sets the flow label value for all packets of a given
flow does not need to store any information about the flow, and any node
that processes the flow label in any way also does not need to store
any information after a packet has been processed. A stateful scenario is
one where a node that sets or processes the flow label value needs to store information about
the flow, including the flow label value. A stateful scenario might also require
a signaling mechanism to establish flow state in the network.
The flow label can be used most simply in stateless scenarios.
This specification concentrates on the stateless model and how it can be used
as a default mechanism. Details of stateful models, signaling, specific flow state
establishment methods and their related service models are out of scope for this
specification. Generic requirements enabling co-existence of different models
are set forth in . The associated
scaling characteristics (such as nodes involved in state establishment,
amount of state maintained by them, and state growth function) will be specific to
particular service models.
The minimum level of IPv6 flow support consists of labeling the
flows. A specific goal is to enable and encourage the use of the
flow label for various forms of stateless load distribution, especially across
Equal Cost Multi-Path (EMCP) and/or Link Aggregation Group (LAG) paths.
ECMP and LAG are methods to bond together multiple physical links used to
procure the required capacity necessary to carry an offered load
greater than the bandwidth of an individual physical link.
IPv6 source nodes SHOULD be able
to label known flows (e.g., TCP connections, application streams),
even if the node itself does not require any flow-specific
treatment. Node requirements for stateless flow
labeling are given in .
This document replaces and Appendix A of .
A rationale for the changes made is documented in .
The present document also includes
a correction to concerning the flow label.
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 .
The 20-bit Flow Label field in the IPv6 header is used by a
node to label packets of a flow. A Flow Label of zero is used to
indicate packets not part of any flow. Packet classifiers can use the
triplet of Flow Label, Source Address, and Destination Address fields
to identify which flow a particular packet belongs to. Packets are
processed in a flow-specific manner by nodes that are able to do so in a
stateless manner, or that have been set up with flow-specific state.
The nature of the specific treatment and the methods for flow state establishment
are out of scope for this specification. However, any node that sets
flow label values according to a stateful scheme MUST ensure that packets
conform to of the present specification if they are sent outside
the network domain using the stateful scheme.
As specified below in , the normal expectation
is that flow label values are uniformly distributed. In this specification,
it is recommended below that a pseudo-random method should be used to achieve such
a uniform distribution. Intentionally, there are no precise mathematical requirements
placed on the distribution or the pseudo-random method.
Once set to a non-zero value, the Flow Label MUST be delivered unchanged to
the destination node(s). A forwarding node MUST NOT change the flow label value in an arriving packet if
it is non-zero. However, there are two qualifications to this rule:
Implementers are advised that forwarding nodes, especially those acting as domain border devices,
might nevertheless be configured to change the flow label value in packets.
This is undetectable, unless some future version of
IPsec authentication protects the flow label value.
To enable stateless load distribution at any point in the
Internet, a network domain should never export packets originating within the domain
whose flow label values do not conform to .
However, neither domain border egress routers nor intermediate
routers/devices (using a flow-label, for example, as a part of an
input-key for a load-distribution hash) can determine by
inspection that a value is not part of a uniform distribution. Therefore, if
nodes within a domain ignore the recommendations of ,
and such packets are forwarded outside the domain, this might result in
undesirable operational implications (e.g., congestion,
reordering) for not only the inappropriately flow-labelled
packets, but also well-behaved flow-labelled packets, during
forwarding at various intermediate devices. Thus, a domain must
protect its peers by never exporting inappropriately labelled
packets originating within the domain. This is why nodes using
a stateful scheme must not set the flow label to a non-zero and
non-uniformly distributed value if the packet will leave their domain.
If it is known to a border router that flow labels originated within the
domain are not uniformly distributed, it will need to
set outgoing flow labels in the same manner as described
for forwarding nodes in .
There is no way to verify whether a flow label has been modified en route
or whether it belongs to a uniform distribution.
Therefore, no Internet-wide mechanism can depend mathematically on immutable and uniformly
distributed flow labels; they have a "best effort" quality. This leads to the following formal rules:
Implementers should be aware that the flow label is an
unprotected field that could have been accidentally
or intentionally changed en route. Implementations MUST
take appropriate steps to protect themselves from being
vulnerable to denial of service and other types of attack that
could result (see ).
Forwarding nodes such as routers and load balancers MUST NOT depend only on Flow Label
values being uniformly distributed. In any usage such as a hash key for load distribution,
the Flow Label bits MUST be combined at least with bits from other sources within the packet,
so as to produce a constant hash value for each flow and a suitable distribution of hash
values across flows.
Although uniformly distributed flow
label values are recommended below, and will always be helpful for load balancing, it is unsafe to assume
their presence in the general case, and the use case needs to work even if the flow label
value is zero.
The use of the Flow Label field does not necessarily signal any
requirement on packet reordering. Especially, the zero label does
not imply that significant reordering is acceptable.
An IPv6 node that does not set the flow label to a non-zero value, or make use of it in any way, MUST
ignore it when receiving or forwarding a packet.
This section defines the minimum requirements for stateless methods of setting the flow label value.
To enable Flow Label based classification, source nodes SHOULD assign
each unrelated transport connection and application data stream to a
new flow. A typical definition of a flow for this purpose is any set
of packets carrying the same 5-tuple {dest addr, source addr, protocol, dest port, source port}.
It is desirable that flow label values should be uniformly distributed
to assist load distribution. It is therefore RECOMMENDED that source hosts support the flow label by
setting the flow label field for all packets of a given flow to the same uniformly distributed pseudo-random value.
Both stateful and stateless methods of assigning a pseudo-random value could be used,
but it is outside the scope of this specification to mandate an algorithm. In a stateless mechanism,
the algorithm SHOULD ensure that the resulting flow label values are unique
with high probability.
An OPTIONAL algorithm for generating such a pseudo-random value is
described in .
[[ NOTE TO RFC EDITOR: The preceding sentence should be deleted, and the reference should be
changed to Informative, if the cited draft is not on the standards track at the time of publication. ]]
A source node which does not otherwise set the flow label
MUST set its value to zero.
A node that forwards a flow whose flow label value in arriving packets is zero
MAY set the flow label value. In that case, it is RECOMMENDED
that the forwarding node sets the flow label field for a flow to a uniformly distributed pseudo-random value.
The same considerations apply as to source hosts setting the flow label; in particular,
the normal case is that a flow is defined by the 5-tuple.
This option, if implemented, would presumably be used by first-hop or ingress routers. It might place a
considerable per-packet processing load on them, even if they adopted a stateless method of
flow identification and label assignment. This is why the principal recommendation is that
the source host should set the label.
The preceding rules taken together allow a given network domain to
include routers that set flow labels on behalf of hosts that do not do so.
They also recommend that flow labels exported
to the Internet are always either zero or uniformly distributed.
This section defines the minimum requirements for stateful methods of setting the flow label value.
The node that sets the flow label MAY also take part in flow state
establishment methods that result in assigning specific treatments to
specific flows, possibly including signaling.
In this case, unlike the stateless case, a source node MUST ensure
that it does not unintentionally reuse Flow
Label values it is currently using or has recently used when creating
new flows. Flow Label values previously used with a specific pair of
source and destination addresses MUST NOT be assigned to new flows
with the same address pair within 120 seconds of the termination of
the previous flow.
To avoid accidental Flow Label value reuse, the source node SHOULD
select new Flow Label values in a well-defined way
and use an initial value that avoids
reuse of recently used Flow Label values each time the system
restarts. The initial value SHOULD be derived from a previous value
stored in non-volatile memory, or in the absence of such history, a
randomly generated initial value using techniques that produce good
randomness properties SHOULD be used.
To enable stateful flow-specific treatment, flow state needs to be established
on all or a subset of the IPv6 nodes on the path from the source to
the destination(s). The methods for the state establishment, as well
as the models for flow-specific treatment will be defined in separate
specifications.
In stateful mechanisms, nodes keeping dynamic flow state MUST NOT assume packets arriving 120
seconds or more after the previous packet of a flow still belong to
the same flow, unless a flow state establishment method in use
defines a longer flow state lifetime or the flow state has been
explicitly refreshed within the lifetime duration.
To enable co-existence of different methods in IPv6 nodes, the
methods MUST meet the following basic requirements:
The method MUST provide the means for flow state clean-up from
the IPv6 nodes providing the flow-specific treatment. Signaling
based methods where the source node is involved are free to
specify flow state lifetimes longer than the default 120
seconds.
Flow state establishment methods MUST be able to recover from
the case where the requested flow state cannot be supported.
reduced the size of the flow label field from 24 to 20 bits.
The references to a 24 bit flow label field on pages 87 and 88 of are updated accordingly.
This section considers security issues raised by the use of the Flow
Label, primarily the potential for denial-of-service attacks, and the
related potential for theft of service by unauthorized traffic
(). addresses the use of the Flow Label in
the presence of IPsec including its interaction with IPsec tunnel
mode and other tunneling protocols. We also note that inspection of
unencrypted Flow Labels may allow some forms of traffic analysis by
revealing some structure of the underlying communications. Even if
the flow label were encrypted, its presence as a constant value in a
fixed position might assist traffic analysis and cryptoanalysis.
The flow label is not protected in any way and can be forged by an on-path
attacker. On the other hand, a uniformly distributed pseudo-random flow label cannot be readily
guessed by an off-path attacker; see
for further discussion.
Since the mapping of network traffic to flow-specific treatment is
triggered by the IP addresses and Flow Label value of the IPv6
header, an adversary may be able to obtain better service by
modifying the IPv6 header or by injecting packets with false
addresses and/or labels. Taken to its limits, such theft-of-service
becomes a denial-of-service attack when the modified or injected
traffic depletes the resources available to forward it and other
traffic streams. A curiosity is that if a DoS attack were undertaken
against a given Flow Label (or set of Flow Labels), then traffic
containing an affected Flow Label might well experience worse-than-
best-effort network performance.
Note that since the treatment of IP headers by nodes is typically
unverified, there is no guarantee that flow labels sent by a node are
set according to the recommendations in this document. Therefore,
any assumptions made by the network about header fields such as flow
labels should be limited to the extent that the upstream nodes are
explicitly trusted.
Since flows are identified by the 3-tuple of the Flow Label and the
Source and Destination Address, the risk of theft or denial of
service introduced by the Flow Label is closely related to the risk
of theft or denial of service by address spoofing. An adversary who
is in a position to forge an address is also likely to be able to
forge a label, and vice versa.
There are two issues with different properties: Spoofing of the Flow
Label only, and spoofing of the whole 3-tuple, including Source and
Destination Address.
The former can be done inside a node which is using or transmitting
the correct source address. The ability to spoof a Flow Label
typically implies being in a position to also forge an address, but
in many cases, spoofing an address may not be interesting to the
spoofer, especially if the spoofer's goal is theft of service, rather
than denial of service.
The latter can be done by a host which is not subject to ingress
filtering or by an intermediate router. Due to its
properties, such is typically useful only for denial of service. In
the absence of ingress filtering, almost any third party could
instigate such an attack.
In the presence of ingress filtering, forging a non-zero Flow Label
on packets that originated with a zero label, or modifying or
clearing a label, could only occur if an intermediate system such as
a router was compromised, or through some other form of man-in-the-
middle attack. However, the risk is limited to traffic receiving
better or worse quality of service than intended. For example, if
Flow Labels are altered or cleared at random, flow classification
will no longer happen as intended, and the altered packets will
receive default treatment. If a complete 3-tuple is forged, the
altered packets will be classified into the forged flow and will
receive the corresponding quality of service; this will create a
denial of service attack subtly different from one where only the
addresses are forged. Because it is limited to a single flow
definition, e.g., to a limited amount of bandwidth, such an attack
will be more specific and at a finer granularity than a normal
address-spoofing attack.
Since flows are identified by the complete 3-tuple, ingress filtering
will, as noted above, mitigate part of the risk. If the
source address of a packet is validated by ingress filtering, there
can be a degree of trust that the packet has not transited a
compromised router, to the extent that ISP infrastructure may be
trusted. However, this gives no assurance that another form of
man-in-the-middle attack has not occurred.
A man-in-the-middle denial of service attack specifically directed
at flow label handling would involve setting unusual flow labels.
For example, an attacker could set all flow labels reaching a given router
to the same arbitrary non-zero value, or could perform rapid cycling of
flow label values such that the packets of a given flow will each have
a different value. Either of these attacks would cause a stateless load
distribution algorithm to perform badly and would cause a stateful
mechanism to behave incorrectly. For this reason, stateless mechanisms
should not use the flow label alone to control load distribution,
and stateful mechanisms should include explicit methods to detect
and ignore suspect flow label values.
The IPsec protocol, as defined in , ,
does not include
the IPv6 header's Flow Label in any of its cryptographic calculations
(in the case of tunnel mode, it is the outer IPv6 header's Flow Label
that is not included). Hence modification of the Flow Label by a
network node has no effect on IPsec end-to-end security, because it
cannot cause any IPsec integrity check to fail. As a consequence,
IPsec does not provide any defense against an adversary's
modification of the Flow Label (i.e., a man-in-the-middle attack).
IPsec tunnel mode provides security for the encapsulated IP header's
Flow Label. A tunnel mode IPsec packet contains two IP headers: an
outer header supplied by the tunnel ingress node and an encapsulated
inner header supplied by the original source of the packet. When an
IPsec tunnel is passing through nodes performing flow classification,
the intermediate network nodes operate on the Flow Label in the outer
header. At the tunnel egress node, IPsec processing includes
removing the outer header and forwarding the packet (if required)
using the inner header. The IPsec protocol requires that the inner
header's Flow Label not be changed by this decapsulation processing
to ensure that modifications to label cannot be used to launch theft-
or denial-of-service attacks across an IPsec tunnel endpoint. This
document makes no change to that requirement; indeed it forbids
changes to the Flow Label.
When IPsec tunnel egress decapsulation processing includes a
sufficiently strong cryptographic integrity check of the encapsulated
packet (where sufficiency is determined by local security policy),
the tunnel egress node can safely assume that the Flow Label in the
inner header has the same value as it had at the tunnel ingress node.
This analysis and its implications apply to any tunneling protocol
that performs integrity checks. Of course, any Flow Label set in an
encapsulating IPv6 header is subject to the risks described in the
previous section.
The Flow Label does nothing to eliminate the need for packet
filtering based on headers past the IP header, if such filtering is
deemed necessary for security reasons on nodes such as firewalls or
filtering routers.
However, security devices that clear or rewrite non-zero flow label values would
be in violation of this specification.
This document requests no action by IANA.
Steve Deering and Alex Conta were co-authors of RFC 3697, on which this document is based.
Valuable comments and contributions were made by
Fred Baker,
Steve Blake,
Remi Despres,
Alan Ford,
Fernando Gont,
Brian Haberman,
Tony Hain,
Joel Halpern,
Qinwen Hu,
Chris Morrow,
Thomas Narten,
Mark Smith,
Pascal Thubert,
Iljitsch van Beijnum,
and other participants in the 6man working group.
Contributors to the development of RFC 3697 included
Ran Atkinson, Steve Blake, Jim Bound, Francis Dupont,
Robert Elz, Tony Hain, Robert Hancock, Bob Hinden, Christian Huitema,
Frank Kastenholz, Thomas Narten, Charles Perkins, Pekka Savola,
Hesham Soliman, Michael Thomas, Margaret Wasserman, and Alex Zinin.
This document was produced using the xml2rfc tool
.
draft-ietf-6man-flow-3697bis-01: update after resolving 11 initial issues, 2011-02-26
draft-ietf-6man-flow-3697bis-00: original version, built from RFC3697 and draft-ietf-6man-flow-update-01, 2011-01-31
&RFC2460;
&RFC2119;
&RFC2205;
&DRAFT-gont;
&RFC2629;
&RFC2827;
&RFC3697;
&RFC4301;
&RFC4302;
&RFC4303;
&DRAFT-rationale;