Network Working Group Michael H. Behringer
Internet Draft Cisco Systems, Inc.
<draft-behringer-mpls-security-04.txt>
Category: Informational
May 2003
Expires: November 2003
Analysis of the Security of BGP/MPLS IP VPNs
Status of this Memo
This document is an Internet-Draft and is in full conformance
with all provisions of Section 10 of RFC2026.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
Abstract
This document analyses the security of the BGP/MPLS IP VPN architecture
as described in [RFC2547bis], especially in comparison with other VPN
technologies such as ATM and Frame Relay. The target audience is
service providers and VPN users. The document consists of two main
parts: First the requirements for security in VPN services are defined,
second BGP/MPLS IP VPNs are examined with respect to these
requirements.
The analysis shows that BGP/MPLS IP VPN networks can be equally secured
as traditional layer-2 VPN networks such as ATM and Frame Relay.
Internet Draft Security of the MPLS Architecture May 2003
Table of Contents
1. Scope and Introduction.............................................2
2. Security Requirements of VPN Networks..............................3
2.1 Address Space, Routing and Traffic Separation...................3
2.2 Hiding of the Core Infrastructure...............................4
2.3 Resistance to Attacks...........................................4
2.4 Impossibility of Label Spoofing.................................5
3. Analysis of BGP/MPLS IP VPN Security...............................5
3.1 Address Space, Routing and Traffic Separation...................5
3.2 Hiding of the BGP/MPLS IP VPN Core Infrastructure...............7
3.3 Resistance to Attacks...........................................8
3.4 Label Spoofing.................................................10
3.5 Comparison with ATM/FR VPNs....................................11
4. Security of advanced BGP/MPLS IP VPN architectures................11
4.1 Carriers' Carrier (CsC)........................................12
4.2 Inter-provider backbones.......................................13
5. What BGP/MPLS IP VPNs Do Not Provide..............................15
5.1 Protection against Misconfigurations of the Core and Attacks
"within" the Core..................................................15
5.2 Data Encryption, Integrity and Origin Authentication...........16
5.3 Customer Network Security......................................16
6. Layer 2 security considerations...................................17
7. Summary and Conclusions...........................................18
Acknowledgements.....................................................19
Author's Address.....................................................19
References...........................................................19
1. Scope and Introduction
As MPLS (multi protocol label switching) is becoming a more wide-spread
technology for providing IP VPN (virtual private network) services, the
security of the BGP/MPLS IP VPN architecture is of increasing concern
to service providers and VPN customers. This document gives an overview
of the security of the BGP/MPLS IP VPN architecture as described in
[RFC2547bis] for both service providers and MPLS users, and compares it
with traditional layer-2 services such as ATM or Frame Relay from a
security perspective.
The term "MPLS core" is defined for this document as the set of PE and
P routers which are used to provide an BGP/MPLS IP VPN service,
typically under the control of a single service provider. This document
assumes that the MPLS core network is trusted and provided in a secure
manner. Thus it does not address basic security concerns such as
securing the network elements against unauthorised access,
misconfigurations of the core, internal (within the core) attacks and
the likes. Should a customer not wish to assume the service provider
draft-behringer-mpls-security-04.txt page 2
Internet Draft Security of the MPLS Architecture May 2003
network as trusted it becomes necessary to use additional security
mechanisms such as IPsec over the MPLS infrastructure. One way to
implement IPsec over BGP/MPLS is described in [Guichard].
Analysis of the security features of routing protocols is only covered
to the extend where it influences BGP/MPLS IP VPNs. IPsec technology is
also not covered, except to highlight the combination of MPLS VPNs with
IPsec.
The overall security of a system depends on three parts: the
architecture, the implementation, and the operation of the system.
Security issues can exist in either part. This document analyses the
architectural security of BGP/MPLS IP VPNs. It does not cover
implementation issues nor operational issues.
This document is targeted at technical staff of service providers and
enterprises. Knowledge of the basic BGP/MPLS IP VPN architecture as
described in [RFC2547bis] is required to understand this document.
2. Security Requirements of VPN Networks
Both service providers offering any type of VPN services and customers
using them have specific demands for security. Mostly they compare MPLS
based solutions with traditional layer 2 based VPN solutions such as
Frame Relay and ATM, since these are widely deployed and accepted. This
section outlines the security requirements that are typically made in
VPN networks. The following section discusses if and how BGP/MPLS IP
VPNs address these requirements, for both the MPLS core and the
connected VPNs.
2.1 Address Space, Routing and Traffic Separation
Between two non-intersecting layer 3 VPNs of an VPN service it is
assumed that the address space between different VPNs is entirely
independent. This means that for example two non-intersecting VPNs must
be able to both use the 10/8 network without any interference. In
addition traffic from one VPN must never enter another VPN. This
includes separation of routing protocol information, so that also
routing tables are separate per VPN. Specifically:
* Any VPN must be able to use the same address space as any other VPN.
* Any VPN must be able to use the same address space as the MPLS core.
* Traffic from one VPN must never flow to another VPN.
* Routing information, as well as distribution and processing of that
information, for one VPN instance must be independent from any other
VPN instance.
draft-behringer-mpls-security-04.txt page 3
Internet Draft Security of the MPLS Architecture May 2003
* Routing information, as well as distribution and processing of that
information, for one VPN instance must be independent from the core.
From a security point of view the basic requirement is to avoid that
packets destined to a host a.b.c.d within a given VPN reach a host with
the same address in another VPN or the core, or get routed to another
VPN even if this address does not exist there.
2.2 Hiding of the Core Infrastructure
The internal structure of the core network (in the case of MPLS PE and
P elements) should not be visible to outside networks (Internet or any
connected VPN). Whilst a breach of this requirement does not lead to a
security problem itself, many service providers feel that it is
advantageous if the internal addressing and network structure remains
hidden to the outside world. A strong argument is that DoS attacks
against a core router for example are much easier to carry out if an
attacker knows the address. Where addresses are not known, they can be
guessed, but with this attacks become more difficult. Ideally the core
should be as invisible to the outside world as a comparable layer 2
(e.g., frame relay, ATM) infrastructure. Core network elements should
also not be accessible from a VPN.
Note that security should never rely on obscurity, i.e., the hiding of
information. On the contrary services should be equally secure if the
implementation is known. However, there is a strong market perception
that hiding of details is advantageous. This point addresses that
market perception.
2.3 Resistance to Attacks
There are two basic types of attacks: Denial-of-Service (DoS) attacks,
where resources become unavailable to authorised users, and intrusions,
where resources become available to un-authorised users.
For attacks that give unauthorised access to resources (intrusions)
there are two basic ways to protect the network: Firstly, to harden
protocols that could be abused (e.g., telnet to a router), secondly to
make the network as inaccessible as possible. The latter is achieved by
a combination of packet filtering or firewalling and address hiding, as
discussed above.
DoS attacks are easier to execute, since in the simplest case a known
IP address might be enough to attack a machine. This can be done using
normal "allowed" traffic, but higher than normal packet rates, so that
other users cannot access the targeted machine. The only way to be
certain not be vulnerable to this kind of attack is to make sure that
draft-behringer-mpls-security-04.txt page 4
Internet Draft Security of the MPLS Architecture May 2003
machines are not reachable, again by packet filtering and optionally
address hiding.
BGP/MPLS IP VPN networks must provide at least the same level of
protection against both forms of attack as current layer 2 networks.
Note that this document concentrates on protecting the core network
against attacks from the "outside", i.e., the Internet and connected
VPNs. Protection against attacks from the "inside", i.e., if an
attacker has logical or physical access to the core network is not
considered here, since any network can be attacked with access from the
inside.
2.4 Impossibility of Label Spoofing
Assuming the address and traffic separation as discussed above, a
potential attacker might try to gain access to other VPNs by inserting
packets with a label that he does not "own". This could be done from
the outside, i.e., another CE router or from the Internet, or from
within the MPLS core. The latter case (from within the core) will not
be discussed, since the assumption is that the core network is provided
in a secure manner. Should protection against an insecure core be
required it is necessary to run IPsec across the MPLS infrastructure,
at least from CE to CE, since the PEs belong to the core.
Depending on the way several CEs are connected to a PE router, it might
be technically possible to intrude into another VPN that is also
connected on that PE, based on layer 2 attack mechanisms. Examples are
802.1Q - label spoofing, or ATM VPI/VCI spoofing. Layer 2 security
issues will be discussed in section 6.
It is required that VPNs cannot abuse the MPLS label mechanisms or
protocols to gain un-authorised access to other VPNs or the core.
3. Analysis of BGP/MPLS IP VPN Security
In this section the BGP/MPLS IP VPN architecture is analysed with
respect to the security requirements listed above.
3.1 Address Space, Routing and Traffic Separation
BGP/MPLS allows distinct IP VPNs to use the same address space, which
can also be private address space [RFC1918]. This is achieved by adding
a 64 bit route distinguisher (RD) to each IPv4 route, making VPN-unique
addresses also unique in the MPLS core. This "extended" address is also
called a "VPN-IPv4 address". Thus customers of an BGP/MPLS IP VPN
service do not need to change current addressing in their networks.
draft-behringer-mpls-security-04.txt page 5
Internet Draft Security of the MPLS Architecture May 2003
There is only one exception, which is the IP addresses of the PE
routers the CE routers are peering with, in the case of using routing
protocols between CE and PE routers (for static routing between PE and
CE this is not an issue). Routing protocols on the CE routers need to
have configured the address of the peer PE router in the core, to be
able to "talk" to the PE router. This address must be unique from the
CE router's perspective. In an environment where the service provider
manages also the CE routers as CPE, this can be made invisible to the
customer. The address space on the CE-PE link (including the peering PE
address) must be considered as part of the VPN address space. However,
since address space can overlap between VPNs, also the CE-PE link
addressing can overlap between VPNs. (Note that for practical
management considerations SPs typically choose to address all CE-PE
links from a global pool, keeping them unique across the entire core.
The considerations of CE-PE addressing are discussed in detail in
[Guichard2].)
Routing separation between the VPNs can also be achieved. Every PE
router maintains a separate Virtual Routing and Forwarding instance
(VRF) for each connected VPN. Each VRF on the PE router is populated
with routes from one VPN, through statically configured routes or
through routing protocols that run between the PE and the CE router.
Since every VPN results in a separate VRF there will be no
interferences between the VPNs on the PE router.
Across the core to the other PE routers this separation is maintained
by adding unique VPN identifiers in multi-protocol BGP, such as the
route distinguisher. VPN routes are exclusively exchanged by MP-BGP
across the core, and this BGP information is not re-distributed to the
core network but only to the other PE routers, where the information is
kept again in VPN specific VRFs. Thus routing across an BGP/MPLS
network is separate per VPN.
On the data plane traffic separation is achieved by the ingress PE
prepending a VPN-specific label to the packets. The packets with the
VPN labels are sent through the core to the egress PE, where the VPN
label is used to determine the correct VPN.
Given the addressing, routing and traffic separation across an BGP/MPLS
IP VPN core network, it can be assumed that this architecture offers in
this respect the same security as comparable layer-2 VPNs such as ATM
or Frame Relay. It is not possible to intrude from a VPN or the core
into other VPNs through the BGP/MPLS IP VPN network, unless this has
been configured specifically.
draft-behringer-mpls-security-04.txt page 6
Internet Draft Security of the MPLS Architecture May 2003
3.2 Hiding of the BGP/MPLS IP VPN Core Infrastructure
For reasons of security service providers and end-customers do not
normally want their network topology revealed to the outside. This is
done to make attacks more difficult: If an attacker doesn't know the
target he can only guess the IP addresses to attack. Since most DoS
attacks don't provide direct feedback to the attacker it would be
difficult to attack the network. It has to be mentioned specifically
that information hiding as such does not provide security. However, in
the market this is a perceived requirement.
With a known IP address a potential attacker can launch a DoS attack
more easily against that device. So the ideal is to not reveal any
information of the internal network to the outside. This applies
equally to the customer networks as to the core. In practice a number
of additional security measures have to be taken, most of all extensive
packet filtering.
For security reasons it is recommended for any core network - MPLS
based or not - to filter packets from the "outside" (Internet or
connected VPNs) destined to the core infrastructure, where possible.
This makes it very hard to attack the core, although some potentially
desired functionality such as pinging core routers will be lost.
Traceroute across the core still works, since it addresses a
destination outside the core.
MPLS does not reveal unnecessary information to the outside, not even
to customer VPNs. The addressing of the core can be done with private
addresses [RFC1918] or public addresses. Since the interface to the
VPNs as well as potentially to the Internet is BGP, there is no need to
reveal any internal information. The only information required in the
case of a routing protocol between PE and CE is the address of the PE
router. If this is not desired, and if no dynamic routing protocol is
required, static routing on unnumbered interfaces can be configured
between the PE and CE. With this measure the BGP/MPLS IP VPN core can
be kept completely hidden.
Customer VPNs will have to advertise their routes as a minimum to the
BGP/MPLS IP VPN core (dynamically or statically), to ensure
reachability across their VPN. Whilst this could be seen as "too open",
the following has to be noted: Firstly, the information known to the
core is not about specific hosts, but networks (routes); this offers
some degree of abstraction. Secondly, in a VPN-only BGP/MPLS IP VPN
network (i.e., no shared Internet access) this is equal to existing
layer-2 models, where the customer has to trust the service provider to
some degree. Also in a FR or ATM network routing information about the
VPNs can be seen on the core network.
draft-behringer-mpls-security-04.txt page 7
Internet Draft Security of the MPLS Architecture May 2003
In a VPN service with shared Internet access the service provider will
typically announce the routes of customers that wish to use the
Internet to his upstream or peer providers. This can be done via a NAT
function to further obscure the addressing information of the
customers' networks. In this case the customer does not reveal more
information to the general Internet than with a general Internet
service. Core information will still not be revealed at all, except for
the peering address(es) of the PE router(s) that hold(s) the peering
with the Internet.
In summary, in a pure MPLS-VPN service, where no Internet access is
provided, the information hiding is as good as on a comparable FR or
ATM network: No addressing information is revealed to third parties or
the Internet. If a customer chooses to access the Internet via the
BGP/MPLS IP VPN core he will have to reveal the same addressing
structure as for a normal Internet service. NAT can be used for further
address hiding. Being reachable from the Internet automatically exposes
a customer network to additional security threats. Appropriate security
mechanisms have to be deployed such as firewalls and intrusion
detection systems. But this is true for any Internet access, over MPLS
or direct.
If a BGP/MPLS IP VPN network has no interconnections to the Internet,
the security is equal to FR or ATM VPN networks. With an Internet
access from the MPLS cloud the service provider has to reveal at least
one IP address (of the peering PE router) to the next provider, and
thus the outside world.
3.3 Resistance to Attacks
Section 3.1 shows that it is not possible to directly intrude into
other VPNs. Another possibility is to attack the MPLS core, and try to
attack other VPNs from there. As shown above it is not possible to
address a P router directly. The only reachable address from a VPN or
the Internet are the peering addresses of the PE routers. Thus there
are two basic ways the BGP/MPLS IP VPN core can be attacked:
1. By attacking the PE routers directly.
2. By attacking the signaling mechanisms of MPLS (mostly routing)
To attack an element of an BGP/MPLS IP VPN network it is first
necessary to know this element, that is, its address. As discussed in
section 3.2 the addressing structure of the BGP/MPLS IP VPN core is
hidden to the outside world. Thus an attacker does not know the IP
address of any router in the core that he wants to attack. The attacker
could now guess addresses and send packets to these addresses. However,
draft-behringer-mpls-security-04.txt page 8
Internet Draft Security of the MPLS Architecture May 2003
due to the address separation of MPLS each incoming packet will be
treated as belonging to the address space of the customer. Thus it is
impossible to reach an internal router, even through IP address
guessing. There is only one exception to this rule, which is the peer
interface of the PE router. This address of the PE is the only attack
point from the outside (a VPN or Internet).
The routing between a VPN and the BGP/MPLS IP VPN core can be
configured two ways:
1. Static; in this case the PE routers are configured with static
routes to the networks behind each CE, and the CEs are configured to
statically point to the PE router for any network in other parts of
the VPN (mostly a default route). There are now two sub-cases: The
static route can point to the IP address of the PE router, or to an
interface of the CE router (e.g., serial0).
2. Dynamic; here a routing protocol (e.g., RIP, OSPF, BGP) is used
exchange the routing information between the CE and the PE at each
peering point.
In the case of a static route from the CE router to the PE router,
which points to an interface, the CE router doesn't need to know any IP
address of the core network, not even of the PE router. This has the
disadvantage of a more extensive (static) configuration, but from a
security point of view is preferable to the other cases. It is now
possible to configure packet filters on the PE interface to deny any
packet to the PE interface. This way the router and the whole core
cannot be attacked.
In all other cases, each CE router needs to know at least the router ID
(RID; peer IP address) of the PE router in the core, and thus has a
potential destination for an attack. One could imagine various attacks
on various services running on a router. In practice access to the PE
router over the CE-PE interface can be limited to the required routing
protocol by using ACLs (access control lists). This limits the point of
attack to one routing protocol, for example BGP. A potential attack
could be to send an extensive number of routes, or to flood the PE
router with routing updates. Both could lead to a denial-of-service,
however, not to unauthorised access.
To restrict this risk it is necessary to configure the routing protocol
on the PE router as securely as possible. This can be done in various
ways:
draft-behringer-mpls-security-04.txt page 9
Internet Draft Security of the MPLS Architecture May 2003
* By ACL, allow the routing protocol only from the CE router, not from
anywhere else. Furthermore, no access other than that should be
allowed to the PE router in the inbound ACL on each CE interface.
* Where available, configure MD-5 authentication for routing
protocols. This is available for BGP [RFC2385], OSPF [RFC2154] and
RIP2 [RFC2082] for example. It avoids that packets could be spoofed
from other parts of the customer network than the CE router. Note
that this requires service provider and customer to agree on a shared
secret between all CE and PE routers. Note that it is necessary to do
this for all VPN customers, it is not sufficient to do this for the
customer with the highest security requirements.
* To configure where available parameters of the routing protocol, to
further secure this communication. For example the rate of routing
updates should be restricted where possible (in BGP this can be done
through dampening). Also, a maximum number of routes accepted per VRF
should be configured where possible.
In summary, it is not possible to intrude from one VPN into other VPNs,
or the core. However, it is theoretically possible to exploit the
routing protocol to execute a DoS attack against the PE router. This in
turn might have negative impact on other VPNs on this PE router. For
this reason PE routers must be extremely well secured, especially on
their interfaces to the CE routers. ACLs must be configured to limit
access only to the port(s) of the routing protocol, and only from the
CE router. MD5 authentication in routing protocols should be used on
all PE-CE peerings. With all these security measures the only possible
attack is a DoS attack against the routing protocol itself. However,
BGP for example has a number of counter-meassures such as prefix
filtering and dampening built into the protocol, to assure stability.
It is also easily possible to track the source of such a potential DoS
attack. Without dynamic routing between CEs and PEs the security is
equivalent to the security of ATM or Frame Relay networks.
3.4 Label Spoofing
Within the MPLS network packets are not forwarded based on the IP
destination address, but based on labels that are pre-pended to the IP
packets by the inbound PE routers. Similar to IP spoofing attacks,
where an attacker replaces the source or destination IP address of a
packet, it is also theoretically possible to spoof the label of an MPLS
packet. In the first section the assumption was made that the core
network is trusted. If this assumption cannot be made IPsec must be run
over the MPLS cloud. Thus in this section the emphasis is on whether it
is possible to insert packets with (spoofed) labels into the MPLS
draft-behringer-mpls-security-04.txt page 10
Internet Draft Security of the MPLS Architecture May 2003
network from the outside, i.e., from a VPN (CE router) or from the
Internet.
Principally the interface between any CE router and its peering PE
router is an IP interface, i.e., without labels. The CE router is
unaware of the MPLS core, and thinks it is sending IP packets to a
simple router. The "intelligence" is done in the PE device, where based
on the configuration, the label is chosen and pre-pended to the packet.
This is the case for all PE routers, towards CE routers as well as the
upstream service provider. All interfaces into the MPLS cloud only
require IP packets, without labels.
For security reasons a PE router should never accept a packet with a
label from a CE router. [RFC3031] specifies: "Therefore, when a labeled
packet is received with an invalid incoming label, it MUST be
discarded, UNLESS it is determined by some means (not within the scope
of the current document) that forwarding it unlabeled cannot cause any
harm." Since accepting labels on the CE interface would allow passing
packets to other VPNs it is not permitted by the RFC.
Thus it is impossible for an outside attacker to send labelled packets
into the BGP/MPLS IP VPN core.
There remains the possibility to spoof the IP address of a packet that
is being sent to the MPLS core. However, since there is strict
addressing separation within the PE router, and each VPN has its own
VRF, this can only do harm to the VPN the spoofed packet originated
from, in other words, a VPN customer can attack himself. MPLS doesn't
add any security risk here.
The Inter-AS and CsC cases are special cases, since on the interfaces
between providers typically packets with labels are exchanged. See
section 4 for an analysis of these architectures.
3.5 Comparison with ATM/FR VPNs
ATM and FR VPN services often enjoy a very high reputation in terms of
security. Although ATM and FR VPNs can also be provided in a secure
manner, it has been reported that also these technologies can have
severe security vulnerabilities [DataComm]. Also in ATM/FR the security
depends on the configuration of the network being secure, and errors
can also lead to security problems.
4. Security of advanced BGP/MPLS IP VPN architectures.
The BGP/MPLS IP VPN architecture as described in [RFC2547] defines the
PE-CE interface as the only external interface, as seen from the
draft-behringer-mpls-security-04.txt page 11
Internet Draft Security of the MPLS Architecture May 2003
service provider network. In this case, the PE treats the CE as
untrusted, and only accepts pure IP packets from the CE. The IP range
however is treated as belonging to the VPN of the CE, thus the PE
maintains full control over VPN separation.
Subsequently, [RFC2547bis] has defined more complex architectures, with
more open interfaces. These interfaces allow the exchange of label
information and labelled packets to and from devices outside the
control of the service provider. This section discusses the security
implications of these architectures.
4.1 Carriers' Carrier (CsC)
In the CsC architecture the CE is linked to a VRF on the PE. The CE may
send labeled packets to the PE. The label has been previously assigned
by the PE to the CE, and represents the LSP from this CE to the remote
CE via the carrier's network.
RFC2547bis specifies for this case: "When the PE receives a labeled
packet from a CE, it must verify that the top label is one that was
distributed to that CE." This ensures that the CE can only use labels
that the PE correctly associates with the corresponding VPN. Packets
with incorrect labels will be discarded, and thus label spoofing is not
possible.
The use of label-maps on the PE equally leaves the control of the label
information entirely with the PE, so that this has no impact on the
security of the solution.
The packet underneath the top label will - as in standard RFC2547
networks - remain local to the carrier's VPN and not be looked at in
the carriers' carrier core. Consequently potential spoofing of
subsequent labels or IP addresses remains also local to the carrier's
VPN, and has no implication on the carriers' carrier core, nor on other
VPNs on that core. This is specifically stated in RFC2547bis in section
6.
Note that if the PE and CE are interconnected using a shared layer 2
infrastructure such as a switch, attacks are possible on layer 2, which
might enable a third party on the shared layer 2 network to intrude
into a VPN on that PE router. RFC2547bis specifies therefore that
either all devices on a shared layer 2 network have to be part of the
same VPN, or the layer 2 network must be split logically to avoid this
issue. This will be discussed in more detail in section 6.
In the CsC architecture the carrier needs to trust the carriers'
carrier for correct configuration and operation. The customer of the
draft-behringer-mpls-security-04.txt page 12
Internet Draft Security of the MPLS Architecture May 2003
carrier thus implicitely needs to trust both his carrier and the
carriers' carrier.
In summary, a correctly configured carriers' carrier network provides
the same level of security as comparable layer 2 networks, or
traditional RFC2547 networks.
4.2 Inter-provider backbones
RFC2547bis specifies three sub-cases for the inter-provider backbone
(Inter-AS) case.
a) VRF-to-VRF connections at the AS border routers
In this case each PE sees and treats the other PE as a CE; each will
not accept labelled packets, and there is no signalling between the PEs
other than inside the VRFs on both sides. Thus the separation of the
VPNs on both sides and the security of those are the same as on a
single AS RFC2547 network. This has already been shown above to have
the same security properties as traditional layer 2 VPNs.
This solution has potential scalability issues in that the ASBRs need
to maintain a VRF per VPN, and all of the VRFs need to hold all routes
of the specific VPNs. Thus an ASBR can run into memory problems
affecting all VPNs if one single VRF contains too many routes. Thus the
service providers needs to assure that the ASBRs are properly
dimensioned, and apply appropriate security meassures such as limiting
the number of routes per VRF.
The two service providers connecting their VPNs in this way must trust
each other. Since the VPNs are physically separated on different (sub-
)interfaces all signalling between ASBRs remains within a given VPN.
This means that no dynamic cross-VPN security breaches are possible.
However, it is conceivable that a service provider connects a specific
connection from a given VPN to a wrong interface, thus interconnecting
two VPNs that should not be connected. This has to be controlled
operationally.
b) EBGP redistribution of labeled VPN-IPv4 routes
from AS to neighboring AS
In this case the ASBRs on both sides hold the full routing information
for all VPNs on both sides, but not in separate VRFs, but in the BGP
database. (Note this is typically limited to the Inter-AS VPNs through
filtering.) The separation inside the PE is maintained through the use
draft-behringer-mpls-security-04.txt page 13
Internet Draft Security of the MPLS Architecture May 2003
of VPN-IPv4 addresses. The control plane between the ASBRs is using MP-
eBGP, exchanging the VPN routes as VPN-IPv4 addresses, and their own
addresses as BGP next hop IPv4 addresses, plus labels to be used on the
data plane.
The data plane is separated through the use of a single label,
representing a VRF or a subset thereof. RFC2547bis states that an ASBR
should only accept packets with a label that it has assigned to this
router. This prevents the insertion of packets with unknown labels, but
it is possible for a service provider to use any label that the ASBR of
the other provider has passed on to the other ASBR. This allows one
provider to insert packets into any VPN of the other provider to which
it has a label.
Also this solution needs to consider the security on layer 2 at the
interconnection. The RFC states that this type of interconnection
should only be implemented on private interconnection points. See
section 6 for more details.
RFC2547bis states for this case that a trust relationship between the
two connecting ASes must exist for this model to work securely.
Effectively all ASes interconnected in this way form together one
single zone of trust. The VPN customer needs to trust all the service
providers involved in this architecture.
c) PEs exchange labeled VPN-IPv4 routes, ASBRs only exchange loopbacks
of PEs with labels.
In this solution there are effectively two control connections between
ASes. The route reflectors (RRs) exchange via multihop eBGP the VPN-
IPv4 routes. The ASBRs only exchange the labeled addresses of those PE
routers that hold VPN routes which are shared between those ASes. This
maintains scalability for the ASBR routers, since they do not need to
know the VPN-IPv4 routes.
In this solution the top label specifies an LSP to an egress PE router,
the second label specifies a VPN connected to this egress PE. The
security of the ASBR connection has the same constraints as in solution
b): An ASBR should only accept packets with top labels that it has
assigned to the other router, thus verifying that the packet is
addressed to a valid PE router. But any label which was assigned to the
other ASBR router will be accepted, thus it is not possible for an ASBR
to distinguish between different egress PEs, nor between different VPNs
on those PEs. A malicious service provider of one AS could therefore
introduce packets into any VPN of the other AS to which it holds valid
information on its ASBR and PEs.
draft-behringer-mpls-security-04.txt page 14
Internet Draft Security of the MPLS Architecture May 2003
This means that such an ASBR-ASBR connection can only be made with a
trusted party, over a private interface, as described in b).
In addition this solution exchanges labeled VPN-IPv4 addresses between
route reflectors (RR) via MP-eBGP. The control plane itself can be
protected via routing authentication [RFC2385], which ensures that the
routing information has been originated by the expected RR and has not
been modified in transit. But the received VPN information cannot be
verified, as in the previous case. The ASes need to trust each other to
configure their respective networks correctly. Again all ASes involved
in this design form together one trusted zone. The customer therefore
needs to trust all the service providers involved.
The difference between case b) and case c) is that in b) the ASBRs act
as iBGP next-hops for their AS, thus each SP needs to know of the other
SP's core only the addresses of the ASBRs. In case c) the SPs exchange
the loopback addresses of their PE routers, thus each SP reveals
information to the other of his PE routers, and these routers must be
accessible from the other AS. As stated above, accessibility does not
necessarily mean insecurity, and networks should never rely on
"security through obscurity". So if the PE routers are appropriately
secured this should not be an issue. However, there is an increasing
perception that network deviced should generally not be accessible.
In addition for case c) scalability considerations, for example for the
number of BGP peerings, have now to be made for the overall network
including all ASes linked this way. So SPs on both sides need to work
together in defining a scalable architecture, probably with route
reflectors.
In summary all of these Inter-AS solutions logically merge several
provider networks together. For all cases of Inter-AS configuration all
ASes together form a single zone of trust, and service providers need
to trust each other. For the VPN customer the security of the overall
solution is equal to the security of traditional RFC2547 networks, but
he needs to trust all service providers involved.
5. What BGP/MPLS IP VPNs Do Not Provide
5.1 Protection against Misconfigurations of the Core and Attacks
"within" the Core
The security mechanisms discussed here assume correct configuration of
the involved network elements on the core network (PE and P routers).
draft-behringer-mpls-security-04.txt page 15
Internet Draft Security of the MPLS Architecture May 2003
Deliberate or inadvertent misconfigurations from SP staff may result in
undesired behaviour including severe security leaks.
Note that this paragraph specifically refers to the core network, i.e.,
the PE and P elements. Misconfiguration of any of the customer side
elements such as the CE router are covered by the security mechanisms
above. This means that a potential attacker must have access to either
PE or P routers to gain advantage from misconfigurations. If an
attacker has access to core elements, or is able to insert into the
core additional equipment, he will be able to attack both the core
network as well as the connected VPNs. Thus the following is important:
* To avoid the risk of misconfigurations it is important that the
equipment is easy to configure, and that SP staff have the
appropriate training and experience when configuring the network.
Also, proper tools are required for configuring the core network.
* To avoid the risk of "internal" attacks the core network must be
properly secured. This includes network element security, management
security, physical security of the service provider infrastructure,
access control to service provider installations and other standard
SP security mechanisms.
BGP/MPLS IP VPNs can only provide a secure service if the core network
is provided in a secure fashion. This document assumes this to be the
case.
There are various approaches to control the security of a core if the
VPN customer cannot or does not want to trust the service provider.
IPsec from customer controlled devices is one of them. [Bonica]
proposes a CE based authentication scheme based on cookies, aimed at
detecting misconfigurations in the MPLS core. [Behringer] proposes a
similar scheme based on using the MD5 routing authentication. Both
schemes aim to detect and prevent misconfigurations in the core.
5.2 Data Encryption, Integrity and Origin Authentication
BGP/MPLS IP VPNs themselves does not provide encryption, integrity or
authentication services. If these are required IPsec should be used
over the MPLS infrastructure. The same applies to ATM and Frame Relay:
Also here IPsec can provide these missing services.
5.3 Customer Network Security
BGP/MPLS IP VPNs can be secured so that they are comparable with other
VPN services. However, the security of the core network is only one
factor for the overall security of a customer's network. Threats in
today's networks do not only come from the "outside" connection, but
draft-behringer-mpls-security-04.txt page 16
Internet Draft Security of the MPLS Architecture May 2003
also from the "inside" and from other entry points (modems for
example). To reach a good security level for a customer network in an
BGP/MPLS infrastructure, MPLS security is necessary but not sufficient.
The same applies to other VPN technologies like ATM or frame relay. See
also [RFC2196] for more information on how to secure a network.
6. Layer 2 security considerations
In most cases of Inter-AS or Carrier's Carrier solutions a network will
be interconnected to other networks via a point-to-point private
connection, that is, a connection which cannot be interfered by third
parties. It is important to understand that the use of any shared
medium layer 2 technology for such interconnections, such as ethernet
switches, may carry addtional security risks.
There are two types of risks involved in a layer 2 infrastructure:
a) Attacks against layer 2 protocols or mechanisms
Risks in a layer 2 environment include many different forms of ARP
attacks, VLAN trunking attacks, or CAM overflow attacks. For example
ARP spoofing allows an attacker to re-direct traffic between two
routers through his device, thus being able to see all packets between
those two routers.
All of those can be prevented by appropriate security meassures, but
often these security concerns are overlooked. It is of utmost
importance that if a shared medium such as a switch is used in the
above scenarios, that all available layer 2 security mechanisms are
used to prevent layer 2 based attacks.
b) Traffic insertion attacks
Where many routers share a common layer 2 network, for example on an
Internet exchange point, it is possible for a third party to introduce
packets into a network. This has been abused in the past on traditional
exchange points by some service providers to default to another
provider on this exchange point. In effect they are sending all their
traffic into the other SPs network, even though the control plane
(routing) might not allow that.
For this reason routers on exchange points or other shared layer 2
connections should only accept non-labelled IP packets into the global
routing table. Any labelled packet must be discarded. This maintains
VPN security of connected networks.
draft-behringer-mpls-security-04.txt page 17
Internet Draft Security of the MPLS Architecture May 2003
However, some of the designs above require the exchange of labelled
packets. This would make it possible for a third party to introduce
labelled packets, which when correctly crafted might be associated with
certain VPNs on an BGP/MPLS IP VPN network, effectively introducing
false packets into a VPN.
The current recommendation is therefore to not accept labelled packets
on generic shared medium layer 2 networks such as Internet exchange
points (IXPs). Where labelled packets are required it is strongly
recommended to use private connections.
7. Summary and Conclusions
BGP/MPLS IP VPNs provide full address and traffic separation as in
traditional layer-2 VPN services. It hides addressing structures of the
core and other VPNs, and it is in today's understanding not possible
from the outside to intrude into the core or other VPNs abusing the
BGP/MPLS mechanisms. It is also not possible to intrude into the MPLS
core if this is properly secured. However, there is a significant
difference between BGP/MPLS based IP VPNs and for example FR or ATM
based VPNs: The control structure of the core is on layer 3 in the case
of MPLS. This caused significant skepticism in the industry towards
MPLS, since this might open the architecture to DoS attacks from other
VPNs or the Internet (if connected).
As shown in this document, it is possible to secure an BGP/MPLS IP VPN
infrastructure to the same level of security than a comparable ATM or
FR service. It is also possible to offer Internet connectivity to MPLS
VPNs in a secure manner, and to interconnect different VPNs via
firewalls. Although ATM and FR services have a strong reputation with
regard to security, it has been shown that also in these networks
security problems can exist [DataComm].
As far as attacks from within the MPLS core are concerned, all VPN
classes (BGP/MPLS, FR, ATM) have the same problem: If an attacker can
install a sniffer, he can read information in all VPNs, and if he has
access to the core devices, he can execute a large number of attacks,
from packet spoofing to introducing a new peer routers. There are a
number of precautions measures outlined above that a service provider
can use to tighten security of the core, but the security of the
BGP/MPLS IP VPN architecture depends on the security of the service
provider. If the service provider is not trusted, the only way to fully
secure a VPN against attacks from the "inside" of the VPN service is to
run IPsec on top, from the CE devices or beyond.
This document discussed many aspects of BGP/MPLS IP VPN security. It
has to be noted explicitly that the overall security of this
draft-behringer-mpls-security-04.txt page 18
Internet Draft Security of the MPLS Architecture May 2003
architecture depends on all components, and is determined by the
security of the weakest part of the solution. For example a perfectly
secured static BGP/MPLS IP VPN network with secured Internet access and
secure management is still open to many attacks if there is a weak
remote access solution in place.
Acknowledgements
The author would like to thank everybody who has provided input to this
document. Specific thanks go to Yakov Rekhter for his continued strong
support, and Loa Andersson, Alexander Manhenke and Jim Guichard for
their extended feedback and support.
Author's Address
Michael H. Behringer
Avda de la Vega, 15
28100 Alcobendas, Madrid
Spain
E-mail: mbehring@cisco.com
References
[Bonica] "CE-to-CE Authentication for Layer 3 VPNs". R. Bonica et al;
draft-ietf-ppvpn-l3vpn-auth-00.txt; work in progress
[Behringer] "MPLS VPN Authentication". M. Behringer, J. Guichard;
draft-behringer-mpls-vpn-auth-00.txt; work in progress
[DataComm] "Frame Relay and ATM: Are they really secure?". Data
Communications Report, Vol 15, No 4, February 2000.
(http://www.yankeegroup.com)
[Guichard] "CE-CE IPSec within an RFC-2547 Network". J. Guichard et al.
draft-guichard-ce-ce-ipsec-00.txt; work in progress
[Guichard2] "Address Allocation for PE-CE links within an RFC2547bis
Network". J. Guichard et al. draft-guichard-pe-ce-addr-00.txt; work in
progress [xxx: ref correct?]
[RFC1918] "Address Allocation for Private Internets". Y. Rekhter et al;
February 1996. (http://search.ietf.org/rfc/rfc1918.txt)
[RFC2082] "RIP-2 MD5 Authentication". F. Baker, R. Atkinson. January
1997. (http://search.ietf.org/rfc/rfc2082.txt)
draft-behringer-mpls-security-04.txt page 19
Internet Draft Security of the MPLS Architecture May 2003
[RFC2154] "OSPF with Digital Signatures". S. Murphy, M. Badger, B.
Wellington. June 1997. (http://search.ietf.org/rfc/rfc2154.txt)
[RFC2196] "Site Security Handbook". B. Fraser. September 1997.
(http://search.ietf.org/rfc/rfc2196.txt)
[RFC2385] "Protection of BGP Sessions via the TCP MD5 Signature
Option". A. Heffernan. August 1998.
(http://search.ietf.org/rfc/rfc2385.txt)
[RFC2547] "BGP/MPLS VPNs". E. Rosen, Y. Rekhter. March 1999.
(http://search.ietf.org/rfc/rfc2547.txt)
[RFC2547bis]: "BGP/MPLS VPNs". E. Rosen, et al. Work in progress.
(draft-ietf-ppvpn-rfc2547bis-03.txt)
[RFC2827] "Network Ingress Filtering: Defeating Denial of Service
Attacks which employ IP Source Address Spoofing". P. Ferguson, D.
Senie. May 2000. (http://search.ietf.org/rfc/rfc2827.txt)
[RFC2828] "Internet Security Glossary". R. Shirey. May 2000.
(http://search.ietf.org/rfc/rfc2828.txt)
[RFC3013] "Recommended Internet Service Provider Security Services and
Procedures". T. Killalea. November 2000.
(http://search.ietf.org/rfc/rfc3013.txt)
[RFC3031] "Multiprotocol Label Switching Architecture". E. Rosen, A.
Viswanathan, R. Callon. January
2001.(http://search.ietf.org/rfc/rfc3031.txt)
Full Copyright Statement
Copyright (C) The Internet Society (2000). All Rights Reserved.
This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain it
or assist in its implementation may be prepared, copied, published
and distributed, in whole or in part, without restriction of any
kind, provided that the above copyright notice and this paragraph
are included on all such copies and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
draft-behringer-mpls-security-04.txt page 20
Internet Draft Security of the MPLS Architecture May 2003
followed, or as required to translate it into languages other than
English.
The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on an
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE."
draft-behringer-mpls-security-04.txt page 21