Internet Engineering Task Force (IETF) F. Gont
Request for Comments: 7707 Huawei Technologies
Obsoletes: 5157 T. Chown
Category: Informational Jisc
ISSN: 2070-1721 March 2016 Network Reconnaissance in IPv6 Networks
IPv6 offers a much larger address space than that of its IPv4
counterpart. An IPv6 subnet of size /64 can (in theory) accommodate
approximately 1.844 * 10^19 hosts, thus resulting in a much lower
host density (#hosts/#addresses) than is typical in IPv4 networks,
where a site typically has 65,000 or fewer unique addresses. As a
result, it is widely assumed that it would take a tremendous effort
to perform address-scanning attacks against IPv6 networks; therefore,
IPv6 address-scanning attacks have been considered unfeasible. This
document formally obsoletes RFC 5157, which first discussed this
assumption, by providing further analysis on how traditional address-
scanning techniques apply to IPv6 networks and exploring some
additional techniques that can be employed for IPv6 network
Status of This Memo
This document is not an Internet Standards Track specification; it is
published for informational purposes.
This document is a product of the Internet Engineering Task Force
(IETF). It represents the consensus of the IETF community. It has
received public review and has been approved for publication by the
Internet Engineering Steering Group (IESG). Not all documents
approved by the IESG are a candidate for any level of Internet
Standard; see Section 2 of RFC 5741.
Information about the current status of this document, any errata,
and how to provide feedback on it may be obtained at
5.4. Application Participation . . . . . . . . . . . . . . . . 255.5. Inspection of the IPv6 Neighbor Cache and Routing Table . 255.6. Inspection of System Configuration and Log Files . . . . 265.7. Gleaning Information from Routing Protocols . . . . . . . 265.8. Gleaning Information from IP Flow Information Export
(IPFIX) . . . . . . . . . . . . . . . . . . . . . . . . . 265.9. Obtaining Network Information with traceroute6 . . . . . 265.10. Gleaning Information from Network Devices Using SNMP . . 275.11. Obtaining Network Information via Traffic Snooping . . . 276. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 277. Security Considerations . . . . . . . . . . . . . . . . . . . 278. References . . . . . . . . . . . . . . . . . . . . . . . . . 288.1. Normative References . . . . . . . . . . . . . . . . . . 288.2. Informative References . . . . . . . . . . . . . . . . . 29Appendix A. Implementation of a Full-Fledged IPv6 Address-
Scanning Tool . . . . . . . . . . . . . . . . . . . 34A.1. Host-Probing Considerations . . . . . . . . . . . . . . . 34A.2. Implementation of an IPv6 Local Address-Scanning Tool . . 35A.3. Implementation of an IPv6 Remote Address-Scanning Tool . 36
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . 37
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 381. Introduction
The main driver for IPv6 [RFC2460] deployment is its larger address
space [CPNI-IPv6]. This larger address space not only allows for an
increased number of connected devices but also introduces a number of
subtle changes in several aspects of the resulting networks. One of
these changes is the reduced host density (the number of hosts
divided by the number of addresses) of typical IPv6 subnetworks, when
compared to their IPv4 counterparts. [RFC5157] describes how this
significantly lower IPv6 host density is likely to make classic
network address-scanning attacks less feasible, since even by
applying various heuristics, the address space to be scanned remains
very large. RFC 5157 goes on to describe some alternative methods
for attackers to glean active IPv6 addresses and provides some
guidance for administrators and implementors, e.g., not using
sequential addresses with DHCPv6.
With the benefit of more than five years of additional IPv6
deployment experience, this document formally obsoletes RFC 5157. It
emphasizes that while address-scanning attacks are less feasible,
they may, with appropriate heuristics, remain possible. At the time
that RFC 5157 was written, observed address-scanning attacks were
typically across ports on the addresses of discovered servers; since
then, evidence that some classic address scanning is occurring is
being witnessed. This text thus updates the analysis on the
feasibility of address-scanning attacks in IPv6 networks, and it
explores a number of additional techniques that can be employed for
IPv6 network reconnaissance. Practical examples and guidance are
also included in the appendices.
On one hand, raising awareness about IPv6 network reconnaissance
techniques may allow (in some cases) network and security
administrators to prevent or detect such attempts. On the other
hand, network reconnaissance is essential for the so-called
"penetration tests" typically performed to assess the security of
production networks. As a result, we believe the benefits of a
thorough discussion of IPv6 network reconnaissance are twofold.
Section 4 analyzes the feasibility of address-scanning attacks (e.g.,
ping sweeps) in IPv6 networks and explores a number of possible
improvements to such techniques. Appendix A describes how the
aforementioned analysis can be leveraged to produce address-scanning
tools (e.g., for penetration testing purposes). Finally, the rest of
this document discusses a number of miscellaneous techniques that
could be leveraged for IPv6 network reconnaissance.
Throughout this document, we consider that bits are numbered from
left to right, starting at 0, and that bytes are numbered from left
to right, starting at 0.
3. Requirements for the Applicability of Network Reconnaissance
Throughout this document, a number of network reconnaissance
techniques are discussed. Each of these techniques has different
requirements on the side of the practitioner, with respect to whether
they require local access to the target network and whether they
require login access (or similar access credentials) to the system on
which the technique is applied.
The following table tries to summarize the aforementioned
requirements and serves as a cross index to the corresponding
| Technique | Local | Login |
| | access | access |
| Remote Address Scanning (Section 4.2) | No | No |
| Local Address Scanning (Section 4.3) | Yes | No |
| DNS Advertised Hosts (Section 5.1.1) | No | No |
| DNS Zone Transfers (Section 5.1.2) | No | No |
| DNS Brute Forcing (Section 5.1.3) | No | No |
| DNS Reverse Mappings (Section 5.1.4) | No | No |
| Leveraging Local Name Resolution and | Yes | No |
| Service Discovery Services (Section 5.2) | | |
| Public Archives (Section 5.3) | No | No |
| Application Participation (Section 5.4) | No | No |
| Inspection of the IPv6 Neighbor Cache and | No | Yes |
| Routing Table (Section 5.5) | | |
| Inspecting System Configuration and Log | No | Yes |
| Files (Section 5.6) | | |
| Gleaning Information from Routing Protocols | Yes | No |
| (Section 5.7) | | |
| Gleaning Information from IP Flow | No | Yes |
| Information Export (IPFIX) (Section 5.8) | | |
| Obtaining Network Information with | No | No |
| traceroute6 (Section 5.9) | | |
| Gleaning Information from Network Devices | No | Yes |
| Using SNMP (Section 5.10) | | |
| Obtaining Network Information via Traffic | Yes | No |
| Snooping (Section 5.11) | | |
Table 1: Requirements for the Applicability of
Network Reconnaissance Techniques
4. IPv6 Address Scanning
This section discusses how traditional address-scanning techniques
(e.g., "ping sweeps") apply to IPv6 networks. Section 4.1 provides
an essential analysis of how address configuration is performed in
IPv6, identifying patterns in IPv6 addresses that can be leveraged to
reduce the IPv6 address search space when performing IPv6 address-
scanning attacks. Section 4.2 discusses IPv6 address scanning of
remote networks. Section 4.3 discusses IPv6 address scanning of
local networks. Section 4.4 discusses existing IPv6 address-scanning
tools. Section 4.5 provides advice on how to mitigate IPv6 address-
scanning attacks. Finally, Appendix A discusses how the insights
obtained in the following subsections can be incorporated into a
fully fledged IPv6 address-scanning tool.
4.1. Address Configuration in IPv6
IPv6 incorporates two automatic address-configuration mechanisms:
Stateless Address Autoconfiguration (SLAAC) [RFC4862] and Dynamic
Host Configuration Protocol for IPv6 (DHCPv6) [RFC3315]. Support for
SLAAC for automatic address configuration is mandatory, while support
for DHCPv6 is optional -- however, most current versions of general-
purpose operating systems support both. In addition to automatic
address configuration, hosts, typically servers, may employ manual
configuration, in which all the necessary information is manually
entered by the host or network administrator into configuration files
at the host.
The following subsections describe each of the possible configuration
mechanisms/approaches in more detail.
4.1.1. Stateless Address Autoconfiguration (SLAAC)
The basic idea behind SLAAC is that every host joining a network will
send a multicasted solicitation requesting network configuration
information, and local routers will respond to the request providing
the necessary information. SLAAC employs two different ICMPv6
message types: ICMPv6 Router Solicitation and ICMPv6 Router
Advertisement messages. Router Solicitation messages are employed by
hosts to query local routers for configuration information, while
Router Advertisement messages are employed by local routers to convey
the requested information.
Router Advertisement messages convey a plethora of network
configuration information, including the IPv6 prefix that should be
used for configuring IPv6 addresses on the local network. For each
local prefix learned from a Router Advertisement message, an IPv6
address is configured by appending a locally generated Interface
Identifier (IID) to the corresponding IPv6 prefix.
The following subsections describe currently deployed policies for
generating the IIDs used with SLAAC.
22.214.171.124. Interface Identifiers Embedding IEEE Identifiers
The traditional SLAAC IIDs are based on the link-layer address of the
corresponding network interface card. For example, in the case of
Ethernet addresses, the IIDs are constructed as follows:
1. The "Universal" bit (bit 6, from left to right) of the address is
set to 1.
2. The word 0xfffe is inserted between the Organizationally Unique
Identifier (OUI) and the rest of the Ethernet address.
For example, the Media Access Control (MAC) address 00:1b:38:83:88:3c
would lead to the IID 021b:38ff:fe83:883c.
A number of considerations should be made about these identifiers.
Firstly, one 16-bit word (bytes 3-4) of the resulting address always
has a fixed value (0xfffe), thus reducing the search space for the
IID. Secondly, the high-order three bytes of the IID correspond to
the OUI of the network interface card vendor. Since not all possible
OUIs have been assigned, this further reduces the IID search space.
Furthermore, of the assigned OUIs, many could be regarded as
corresponding to legacy devices and thus are unlikely to be used for
Internet-connected IPv6-enabled systems, yet further reducing the IID
search space. Finally, in some scenarios, it could be possible to
infer the OUI in use by the target network devices, yet narrowing
down the possible IIDs even more.
For example, an organization known for being provisioned by vendor
X is likely to have most of the nodes in its organizational
network with OUIs corresponding to vendor X.
These considerations mean that in some scenarios, the original IID
search space of 64 bits may be effectively reduced to 2^24 or n *
2^24 (where "n" is the number of different OUIs assigned to the
Furthermore, if just one host address is detected or known within a
subnet, it is not unlikely that, if systems were ordered in a batch,
they may have sequential MAC addresses. Additionally, given a MAC
address observed in one subnet, sequential or nearby MAC addresses
may be seen in other subnets in the same site.
[RFC7136] notes that all bits of an IID should be treated as
"opaque" bits. Furthermore, [DEFAULT-IIDS] is currently in the
process of changing the default IID generation scheme to align
with [RFC7217] (as described below in Section 126.96.36.199), such that
IIDs are semantically opaque and do not follow any patterns.
Therefore, the traditional IIDs based on link-layer addresses are
expected to become less common over time.
188.8.131.52. Interface Identifiers of Virtualization Technologies
IIDs resulting from virtualization technologies can be considered a
specific subcase of IIDs embedding IEEE identifiers (please see
Section 184.108.40.206): they employ IEEE identifiers, but part of the IID
has specific patterns. The following subsections describe IIDs of
some popular virtualization technologies.
All automatically generated MAC addresses in VirtualBox virtual
machines employ the OUI 08:00:27 [VBox2011]. This means that all
addresses resulting from traditional SLAAC will have an IID of the
form a00:27ff:feXX:XXXX, thus effectively reducing the IID search
space from 64 bits to 24 bits.
220.127.116.11.2. VMware ESX Server
The VMware ESX server (versions 1.0 to 2.5) provides yet a more
interesting example. Automatically generated MAC addresses have the
following pattern [vmesx2011]:
1. The OUI is set to 00:05:69.
2. The next 16 bits of the MAC address are set to the same value as
the last 16 bits of the console operating system's primary IPv4
3. The final 8 bits of the MAC address are set to a hash value based
on the name of the virtual machine's configuration file.
This means that, assuming the console operating system's primary IPv4
address is known, the IID search space is reduced from 64 bits to 8
On the other hand, manually configured MAC addresses in the VMware
ESX server employ the OUI 00:50:56, with the low-order three bytes of
the MAC address being in the range 00:00:00-3F:FF:FF (to avoid
conflicts with other VMware products). Therefore, even in the case
of manually configured MAC addresses, the IID search space is reduced
from 64 bits to 22 bits.
18.104.22.168.3. VMware vSphere
VMware vSphere [vSphere] supports these default MAC address
o Generated addresses
* Assigned by the vCenter server
* Assigned by the ESXi host
o Manually configured addresses
By default, MAC addresses assigned by the vCenter server use the OUI
00:50:56 and have the format 00:50:56:XX:YY:ZZ, where XX is
calculated as (0x80 + vCenter Server ID (in the range 0x00-0x3F)),
and XX and YY are random two-digit hexadecimal numbers. Thus, the
possible IID range is 00:50:56:80:00:00-00:50:56:BF:FF:FF; therefore,
the search space for the resulting SLAAC addresses will be 22 bits.
MAC addresses generated by the ESXi host use the OUI 00:0C:29 and
have the format 00:0C:29:XX:YY:ZZ, where XX, YY, and ZZ are the last
three octets in hexadecimal format of the virtual machine Universally
Unique Identifier (UUID) (based on a hash calculated with the UUID of
the ESXi physical machine and the path to a configuration file).
Thus, the MAC addresses will be in the range
00:0C:29:00:00:00-00:0C:29:FF:FF:FF; therefore, the search space for
the resulting SLAAC addresses will be 24 bits.
Finally, manually configured MAC addresses employ the OUI 00:50:56,
with the low-order three bytes being in the range 00:00:00-3F:FF:FF
(to avoid conflicts with other VMware products). Therefore, the
resulting MAC addresses will be in the range
00:50:56:00:00:00-00:50:56:3F:FF:FF, and the search space for the
corresponding SLAAC addresses will be 22 bits.
22.214.171.124. Temporary Addresses
Privacy concerns [Gont-DEEPSEC2011] [RFC7721] regarding IIDs
embedding IEEE identifiers led to the introduction of "Privacy
Extensions for Stateless Address Autoconfiguration in IPv6"
[RFC4941], also known as "temporary addresses" or "privacy
addresses". Essentially, "temporary addresses" produce random
addresses by concatenating a random identifier to the
autoconfiguration IPv6 prefix advertised in a Router Advertisement
In addition to their unpredictability, these addresses are
typically short-lived, such that even if an attacker were to learn
of one of these addresses, they would be of use for a limited
period of time. A typical implementation may keep a temporary
address preferred for 24 hours, and configured but deprecated for
It is important to note that "temporary addresses" are generated in
addition to the stable addresses [RFC7721] (such as the traditional
SLAAC addresses based on IEEE identifiers): stable SLAAC addresses
are meant to be employed for "server-like" inbound communications,
while "temporary addresses" are meant to be employed for "client-
like" outbound communications. This means that implementation/use of
"temporary addresses" does not prevent an attacker from leveraging
the predictability of stable SLAAC addresses, since "temporary
addresses" are generated in addition to (rather than as a replacement
of) the stable SLAAC addresses (such as those derived from IEEE
The benefit that temporary addresses offer in this context is that
they reduce the exposure of the host addresses to any third parties
that may observe traffic sent from a host where temporary addresses
are enabled and used by default. But, in the absence of firewall
protection for the host, its stable SLAAC address remains liable to
be scanned from off-site.
126.96.36.199. Constant, Semantically Opaque IIDs
In order to mitigate the security implications arising from the
predictable IPv6 addresses derived from IEEE identifiers, Microsoft
Windows produced an alternative scheme for generating "stable
addresses" (in replacement of the ones embedding IEEE identifiers).
The aforementioned scheme is believed to be an implementation of RFC
4941 [RFC4941], but without regenerating the addresses over time.
The resulting IIDs are constant across system bootstraps, and also
constant across networks.
Assuming no flaws in the aforementioned algorithm, this scheme would
remove any patterns from the SLAAC addresses.
However, since the resulting IIDs are constant across networks,
these addresses may still be leveraged for host-tracking purposes
The benefit of this scheme is thus that the host may be less readily
detected by applying heuristics to an address-scanning attack, but,
in the absence of concurrent use of temporary addresses, the host is
liable to be tracked across visited networks.
188.8.131.52. Stable, Semantically Opaque IIDs
In response to the predictability issues discussed in Section 184.108.40.206
and the privacy issues discussed in [RFC7721], the IETF has
standardized (in [RFC7217]) a method for generating IPv6 IIDs to be
used with IPv6 SLAAC, such that addresses configured using this
method are stable within each subnet, but the IIDs change when hosts
move from one subnet to another. The aforementioned method is meant
to be an alternative to generating IIDs based on IEEE identifiers,
such that the benefits of stable addresses can be achieved without
sacrificing the privacy of users.
Implementation of this method (in replacement of IIDs based on IEEE
identifiers) eliminates any patterns from the IID, thus benefiting
user privacy and reducing the ease with which addresses can be
4.1.2. Dynamic Host Configuration Protocol for IPv6 (DHCPv6)
DHCPv6 can be employed as a stateful address configuration mechanism,
in which a server (the DHCPv6 server) leases IPv6 addresses to IPv6
hosts. As with the IPv4 counterpart, addresses are assigned
according to a configuration-defined address range and policy, with
some DHCPv6 servers assigning addresses sequentially, from a specific
range. In such cases, addresses tend to be predictable.
For example, if the prefix 2001:db8::/64 is used for assigning
addresses on the local network, the DHCPv6 server might
(sequentially) assign addresses from the range 2001:db8::1 -
In most common scenarios, this means that the IID search space will
be reduced from the original 64 bits to 8 or 16 bits. [RFC5157]
recommended that DHCPv6 instead issue addresses randomly from a large
pool; that advice is repeated here. [IIDS-DHCPv6] specifies an
algorithm that can be employed by DHCPv6 servers to produce stable
addresses that do not follow any specific pattern, thus resulting in
an IID search space of 64 bits.
4.1.3. Manually Configured Addresses
In some scenarios, node addresses may be manually configured. This
is typically the case for IPv6 addresses assigned to routers (since
routers do not employ automatic address configuration) but also for
servers (since having a stable address that does not depend on the
underlying link-layer address is generally desirable).
While network administrators are mostly free to select the IID from
any value in the range 1 - 2^64, for the sake of simplicity (i.e.,
ease of remembering), they tend to select addresses with one of the
o low-byte addresses: in which most of the bytes of the IID are set
to 0 (except for the least significant byte)
o IPv4-based addresses: in which the IID embeds the IPv4 address of
the network interface (as in 2001:db8::192.0.2.1)
o service port addresses: in which the IID embeds the TCP/UDP
service port of the main service running on that node (as in
2001:db8::80 or 2001:db8::25)
o wordy addresses: which encode words (as in 2001:db8::bad:cafe)
Each of these patterns is discussed in detail in the following
220.127.116.11. Low-Byte Addresses
The most common form of low-byte addresses is that in which all the
bytes of the IID (except the least significant bytes) are set to zero
(as in 2001:db8::1, 2001:db8::2, etc.). However, it is also common
to find similar addresses in which the two lowest-order 16-bit words
(from the right to left) are set to small numbers (as in
2001::db8::1:10, 2001:db8::2:10, etc.). Yet it is not uncommon to
find IPv6 addresses in which the second lowest-order 16-bit word
(from right to left) is set to a small value in the range
0x0000:0x00ff, while the lowest-order 16-bit word (from right to
left) varies in the range 0x0000:0xffff. It should be noted that all
of these address patterns are generally referred to as "low-byte
addresses", even when, strictly speaking, it is not only the lowest-
order byte of the IPv6 address that varies from one address to
In the worst-case scenario, the search space for this pattern is 2^24
(although most systems can be found by searching 2^16 or even 2^8
18.104.22.168. IPv4-Based Addresses
The most common form of these addresses is that in which an IPv4
address is encoded in the lowest-order 32 bits of the IPv6 address
(usually as a result of the address notation of the form
2001:db8::192.0.2.1). However, it is also common for administrators
to encode each of the bytes of the IPv4 address in each of the 16-bit
words of the IID (as in, e.g., 2001:db8::192:0:2:1).
Therefore, the search space for addresses following this pattern is
that of the corresponding IPv4 prefix (or twice the size of that
search space if both forms of "IPv4-based addresses" are to be
22.214.171.124. Service-Port Addresses
Addresses following this pattern include the service port (e.g., 80
for HTTP) in the lowest-order byte of the IID and have the rest of
the bytes of the IID set to zero. There are a number of variants for
this address pattern:
o The lowest-order 16-bit word (from right to left) may contain the
service port, and the second lowest-order 16-bit word (from right
to left) may be set to a number in the range 0x0000-0x00ff (as in,
o The lowest-order 16-bit word (from right to left) may be set to a
value in the range 0x0000-0x00ff, while the second lowest-order
16-bit word (from right to left) may contain the service port (as
in, e.g., 2001:db8::80:1).
o The service port itself might be encoded in decimal or in
hexadecimal notation (e.g., an address embedding the HTTP port
might be 2001:db8::80 or 2001:db8::50) -- with addresses encoding
the service port as a decimal number being more common.
Considering a maximum of 20 popular service ports, the search space
for addresses following this pattern is, in the worst-case scenario,
10 * 2^11.
126.96.36.199. Wordy Addresses
Since the IPv6 address notation allows for a number of hexadecimal
digits, it is not difficult to encode words into IPv6 addresses (as
in, e.g., 2001:db8::bad:cafe).
Addresses following this pattern are likely to be explored by means
of "dictionary attacks"; therefore, computing the corresponding
search space is not straightforward.
4.1.4. IPv6 Addresses Corresponding to Transition/Coexistence
Some transition/coexistence technologies might be leveraged to reduce
the target search space of remote address-scanning attacks, since
they specify how the corresponding IPv6 address must be generated.
For example, in the case of Teredo [RFC4380], the 64-bit IID is
generated from the IPv4 address observed at a Teredo server along
with a UDP port number.
For obvious reasons, the search space for these addresses will depend
on the specific transition/coexistence technology being employed.
4.1.5. IPv6 Address Assignment in Real-World Network Scenarios
Figures 1, 2, and 3 provide a summary of the results obtained by
[Gont-LACSEC2013] when measuring the address patterns employed by web
servers, name servers, and mail servers, respectively. Figure 4
provides a rough summary of the results obtained by [Malone2008] for
IPv6 routers. Figure 5 provides a summary of the results obtained by
[Ford2013] for clients.
| Address type | Percentage |
| IEEE-based | 7.72% |
| Embedded-IPv4 | 14.31% |
| Embedded-Port | 0.21% |
| ISATAP | 1.06% |
| Randomized | 69.73% |
| Low-byte | 6.23% |
| Byte-pattern | 0.74% |
Figure 5: Measured Client Addresses
"ISATAP" stands for "Intra-Site Automatic Tunnel Addressing
It should be clear from these measurements that a very high
percentage of host and router addresses follow very specific
Figure 5 shows that while around 70% of clients observed in this
measurement appear to be using temporary addresses, a significant
number of clients still expose IEEE-based addresses and addresses
using embedded IPv4 (thus also revealing IPv4 addresses). Besides,
as noted in Section 188.8.131.52, temporary addresses are employed along
with stable IPv6 addresses; thus, hosts employing a temporary address
may still be the subject of address-scanning attacks that target
their stable address(es).
[ADDR-ANALYSIS] contains a spatial and temporal analysis of IPv6
addresses corresponding to clients and routers.
4.2. IPv6 Address Scanning of Remote Networks
Although attackers have been able to get away with "brute-force"
address-scanning attacks in IPv4 networks (thanks to the lesser
search space), successfully performing a brute-force address-scanning
attack of an entire /64 network would be infeasible. As a result, it
is expected that attackers will leverage the IPv6 address patterns
discussed in Section 4.1 to reduce the IPv6 address search space.
IPv6 address scanning of remote networks should consider an
additional factor not present for the IPv4 case: since the typical
IPv6 subnet is a /64, scanning an entire /64 could, in theory, lead
to the creation of 2^64 entries in the Neighbor Cache of the last-hop
router. Unfortunately, a number of IPv6 implementations have been
found to be unable to properly handle a large number of entries in
the Neighbor Cache; hence, these address-scanning attacks may have
the side effect of resulting in a Denial-of-Service (DoS) attack
[RFC7421] discusses the "default" /64 boundary for host subnets and
the assumptions surrounding it. While there are reports of sites
implementing IPv6 subnets of size /112 or smaller to reduce concerns
about the above attack, such smaller subnets are likely to make
address-scanning attacks more feasible, in addition to encountering
the issues with non-/64 host subnets discussed in [RFC7421].
4.2.1. Reducing the Subnet ID Search Space
When address scanning a remote network, consideration is required to
select which subnet IDs to choose. A typical site might have a /48
allocation, which would mean up to 65,000 or so IPv6 /64 subnets to
However, in the same way the search space for the IID can be reduced,
we may also be able to reduce the subnet ID search space in a number
of ways, by guessing likely address plan schemes or using any
complementary clues that might exist from other sources or
observations. For example, there are a number of documents available
online (e.g., [RFC5375]) that provide recommendations for the
allocation of address space, which address various operational
considerations, including Regional Internet Registry (RIR) assignment
policy, ability to delegate reverse DNS zones to different servers,
ability to aggregate routes efficiently, address space preservation,
ability to delegate address assignment within the organization,
ability to add/allocate new sites/prefixes to existing entities
without updating Access Control Lists (ACLs), and ability to
de-aggregate and advertise subspaces via various Autonomous System
Address plans might include use of subnets that:
o Run from low ID upwards, e.g., 2001:db8:0::/64, 2001:db8:1::/64,
o Use building numbers, in hexadecimal or decimal form.
o Use Virtual Local Area Network (VLAN) numbers.
o Use an IPv4 subnet number in a dual-stack target, e.g., a site
with a /16 for IPv4 might use /24 subnets, and the IPv6 address
plan may reuse the third byte of the IPv4 address as the IPv6
o Use the service "color", as defined for service-based prefix
coloring, or semantic prefixes. For example, a site using a
specific coloring for a specific service such as Voice over IP
(VoIP) may reduce the subnet ID search space for those devices.
The net effect is that the address space of an organization may be
highly structured, and allocations of individual elements within this
structure may be predictable once other elements are known.
In general, any subnet ID address plan may convey information, or be
based on known information, which may in turn be of advantage to an
4.3. IPv6 Address Scanning of Local Networks
IPv6 address scanning in Local Area Networks (LANs) could be
considered, to some extent, a completely different problem than that
of scanning a remote IPv6 network. The main difference is that use
of link-local multicast addresses can relieve the attacker of
searching for unicast addresses in a large IPv6 address space.
While a number of other network reconnaissance vectors (such as
network snooping, leveraging Neighbor Discovery traffic, etc.) are
available when scanning a local network, this section focuses only
on address-scanning attacks (a la "ping sweep").
An attacker can simply send probe packets to the all-nodes link-local
multicast address (ff02::1), such that responses are elicited from
all local nodes.
Since Windows systems (Vista, 7, etc.) do not respond to ICMPv6 Echo
Request messages sent to multicast addresses, IPv6 address-scanning
tools typically employ a number of additional probe packets to elicit
responses from all the local nodes. For example, unrecognized IPv6
options of type 10xxxxxx elicit Internet Control Message Protocol
version 6 (ICMPv6) Parameter Problem, code 2, error messages.
Many address-scanning tools discover only IPv6 link-local addresses
(rather than, e.g., the global addresses of the target systems):
since the probe packets are typically sent with the attacker's IPv6
link-local address, the "victim" nodes send the response packets
using the IPv6 link-local address of the corresponding network
interface (as specified by the IPv6 address-selection rules
[RFC6724]). However, sending multiple probe packets, with each
packet employing source addresses from different prefixes, typically
helps to overcome this limitation.
4.4. Existing IPv6 Address-Scanning Tools
4.4.1. Remote IPv6 Network Address Scanners
IPv4 address-scanning tools have traditionally carried out their task
by probing an entire address range (usually the entire address range
comprised by the target subnetwork). One might argue that the reason
for which they have been able to get away with such somewhat
"rudimentary" techniques is that the scale or challenge of the task
is so small in the IPv4 world that a "brute-force" attack is "good
enough". However, the scale of the "address-scanning" task is so
large in IPv6 that attackers must be very creative to be "good
enough". Simply sweeping an entire /64 IPv6 subnet would just not be
Many address-scanning tools do not even support sweeping an IPv6
address range. On the other hand, the alive6 tool from [THC-IPV6]
supports sweeping address ranges, thus being able to leverage some
patterns found in IPv6 addresses, such as the incremental addresses
resulting from some DHCPv6 setups. Finally, the scan6 tool from
[IPv6-Toolkit] supports sweeping address ranges and can also leverage
all the address patterns described in Section 4.1 of this document.
Clearly, a limitation of many of the currently available tools for
IPv6 address scanning is that they lack an appropriately tuned
"heuristics engine" that can help reduce the search space, such that
the problem of IPv6 address scanning becomes tractable.
It should be noted that IPv6 network monitoring and management tools
also need to build and maintain information about the hosts in their
network. Such systems can no longer scan internal systems in a
reasonable time to build a database of connected systems. Rather,
such systems will need more efficient approaches, e.g., by polling
network devices for data held about observed IP addresses, MAC
addresses, physical ports used, etc. Such an approach can also
enhance address accountability, by mapping IPv4 and IPv6 addresses to
observed MAC addresses. This of course implies that any access
control mechanisms for querying such network devices, e.g., community
strings for SNMP, should be set appropriately to avoid an attacker
being able to gather address information remotely.
4.4.2. Local IPv6 Network Address Scanners
There are a variety of publicly available local IPv6 network address-
o Current versions of nmap [nmap2015] implement this functionality.
o The Hacker's Choice (THC) IPv6 Attack Toolkit [THC-IPV6] includes
a tool (alive6) that implements this functionality.
o SI6 Network's IPv6 Toolkit [IPv6-Toolkit] includes a tool (scan6)
that implements this functionality.
IPv6 address-scanning attacks can be mitigated in a number of ways.
A non-exhaustive list of the possible mitigations includes:
o Employing [RFC7217] (stable, semantically opaque IIDs) in
replacement of addresses based on IEEE identifiers, such that any
address patterns are eliminated.
o Employing Intrusion Prevention Systems (IPSs) at the perimeter.
o Enforcing IPv6 packet filtering where applicable (see, e.g.,
o Employing manually configured MAC addresses if virtual machines
are employed and "resistance" to address-scanning attacks is
deemed desirable, such that even if the virtual machines employ
IEEE-derived IIDs, they are generated from non-predictable MAC
o Avoiding use of sequential addresses when using DHCPv6. Ideally,
the DHCPv6 server would allocate random addresses from a large
pool (see, e.g., [IIDS-DHCPv6]).
o Using the "default" /64 size IPv6 subnet prefixes.
o In general, avoiding being predictable in the way addresses are
It should be noted that some of the aforementioned mitigations are
operational, while others depend on the availability of specific
protocol features (such as [RFC7217]) on the corresponding nodes.
Additionally, while some resistance to address-scanning attacks is
generally desirable (particularly when lightweight mitigations are
available), there are scenarios in which mitigation of some address-
scanning vectors is unlikely to be a high priority (if at all
possible). And one should always remember that security by obscurity
is not a reasonable defense in itself; it may only be one (relatively
small) layer in a broader security environment.
Two of the techniques discussed in this document for local address-
scanning attacks are those that employ multicasted ICMPv6 Echo
Requests and multicasted IPv6 packets containing unsupported options
of type 10xxxxxx. These two vectors could be easily mitigated by
configuring nodes to not respond to multicasted ICMPv6 Echo Requests
(default on Windows systems) and by updating the IPv6 specifications
(and/or possibly configuring local nodes) such that multicasted
packets never elicit ICMPv6 error messages (even if they contain
unsupported options of type 10xxxxxx).
[SMURF-AMPLIFIER] proposed such an update to the IPv6
In any case, when it comes to local networks, there are a variety of
network reconnaissance vectors. Therefore, even if address-scanning
vectors were mitigated, an attacker could still rely on, e.g.,
protocols employed for the so-called "service discovery protocols"
(see Section 5.2) or eventually rely on network snooping as a last
resort for network reconnaissance. There is ongoing work in the IETF
on extending mDNS, or at least DNS-based service discovery, to work
across a whole site, rather than in just a single subnet, which will
have associated security implications.
In the previous subsections, we have shown why a /64 host subnet may
be more vulnerable to address-based scanning than might intuitively
be thought and how an attacker might reduce the target search space
when performing an address-scanning attack.
We have described a number of mitigations against address-scanning
attacks, including the replacement of traditional SLAAC with stable
semantically opaque IIDs (which requires support from system
vendors). We have also offered some practical guidance in regard to
the principle of avoiding predictability in host addressing schemes.
Finally, examples of address-scanning approaches and tools are
discussed in the appendices.
While most early IPv6-enabled networks remain dual stack, they are
more likely to be scanned and attacked over IPv4 transport, and one
may argue that the IPv6-specific considerations discussed here are
not of an immediate concern. However, an early IPv6 deployment
within a dual-stack network may be seen by an attacker as a
potentially "easier" target if the implementation of security
policies is not as strict for IPv6 (for whatever reason). As
IPv6-only networks become more common, the above considerations will
be of much greater importance.