Network Working Group                                         Alex Zinin
Internet Draft                                                   Alcatel
Expiration Date: August 2003                               February 2003
File name: draft-zinin-rtg-dos-00.txt

            Protecting Internet Routing Infrastructure from
                          Outsider CPU Attacks


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. Internet Drafts may be updated, replaced, or obsoleted by
   other documents at any time. It is not appropriate to use Internet
   Drafts as reference material or to cite them other than as a "working
   draft" or "work in progress".

   The list of current Internet-Drafts can be accessed at

   The list of Internet-Draft Shadow Directories can be accessed at


   The mechanism described in this document helps to secure an Internet
   Service Provider's router infrastructure from outsider attacks,
   including (but not limited to) Distributed denial of service (DDoS)
   attacks based on CPU and/or queue exhaustion (e.g., TCP SYN flooding
   and flooding of invalid MD5-signed routing protocol packets.) The
   presented approach is based on explicitly marking control packets
   from trusted sources by different link-layer encapsulation and does
   not require any modifications to user data exchange protocols, ICMP,
   routing protocols or changes to existing hardware in routers, which
   allows it to be deployed quickly throughout the Internet.

Zinin                                                           [Page 1]

INTERNET DRAFT            Router DoS protection            February 2003

1 Introduction

 1.1 Problem Description

   The packet authentication mechanisms currently used in Internet
   routing protocols [OSPF, TCP-MD5] leave a generic threat open for an
   outside attacker--overloading the control CPUs of the routers with
   packets that look like they belong to a valid routing protocol
   adjacency or a peering session, yet are fake and would be discarded
   because of invalid digest value. Because all IP parameters of valid
   and faked packets look absolutely identical, it is impossible to
   reject faked packets earlier in the process. This leads to
   overloading of internal queues allocated for control traffic (routing
   and signaling protocols), and hence dropping of legitimate control
   packets. This, combined with high CPU utilization, results in
   destruction of routing protocol sessions and finally in denial of
   service by the network. It is interesting to observe that as security
   mechanisms in routing protocols become more sophisticated and
   computationally expensive, it becomes easier for an attacker to mount
   a CPU-exhaustion-based attack against a router.

   Another example of an attack mountable against routers is the simple
   SYN-flood attack, which could potentially exhaust the router's CPU.

   The in-band nature of IP routing and signaling creates a perfect
   environment for an attacker to put the network itself out of service.
   The fundamental problems leading to the possibility of a DoS attack
   on a router are (a) legitimate and forged packets share resources
   inside the router (such as queues) before the authentication check is
   performed, and (b) the negative authentication decision is
   computationally expensive enough to discourage router vendors from
   performing the check at the line rate. In the latter case, it is
   important to note that the lack of line-rate processing significantly
   increases the router's susceptibility to a distributed DoS attack.

 1.2 Existing Approaches and Disadvantages

   Potential approaches to the problem known to date include:

     1.   Adding specialized HW elements to the line-card architecture
          that would allow the line cards to identify packets that need
          to be authenticated (e.g., OSPF, BGP, RSVP) and perform the
          MD5 check at the line rate (before the packets are put in any
          queue), as well as identify TCP SYN packets and limit the rate
          at which they are sent to the control card.

     2.   Perform aggressive packet filtering at the edges of the net-
          work, on both customer-facing and service provider peering

Zinin                                                           [Page 2]

INTERNET DRAFT            Router DoS protection            February 2003

          interfaces to make sure that packets destined for the internal
          routers are not received from outside the network.

     3.   Use a completely separate set of links for control protocols
          and customer data, i.e. out-of-band network control.

   Below are the disadvantages of these methods (correspondingly):

     1.   From the service provider's perspective, additional HW
          increases the cost of the system and requires upgrades of the
          line-cards of all routers in a service provider's network.
          From the Internet security perspective, it will take years
          before a considerable number of service providers upgrade
          their routing infrastructure, and thus before the threat of
          DoS attack on the Internet routing system is sufficiently mit-

     2.   Many of today's deployed Internet core routers do not have the
          ability to perform line-rate access control list (ACL) pro-
          cessing at high speeds, which means that the inter-service
          provider links will remain insecure. Combined with the fact
          that not all service providers filter potentially dangerous
          packets on the customer interfaces, this approach has the same
          disadvantages from the deployment and Internet security per-
          spective as the first approach.

     3.   While the out-of-band control scheme is extremely interesting,
          implementation could require substantial modification to the
          routing protocols and complete re-architecting of the service
          provider networks. From the architectural point of view, it
          would also require a major shift from the assumption that con-
          trol-plane connectivity implies connectivity of the data

   The solution described in this document allows service providers to
   improve their network without major hardware upgrades , changes to
   routing protocols or network architecture, and with limited software

2 Solution

 2.1 Overview

   The proposed mechanism uses the fact that there are only a limited
   number of devices in a network that have a legitimate right to send a
   packet to a router's control plane. This set of devices includes, of
   course, other routers in the service provider's network, and network
   operations center (NOC) machines.  The rest of the devices in the

Zinin                                                           [Page 3]

INTERNET DRAFT            Router DoS protection            February 2003

   Internet, including user hosts, and routers in other service provider
   networks should not need to send packets to the routers internal to
   the first provider's network..

   They key aspect of the proposal is marking of packets from the set of
   trusted devices in a way that it would either be impossible to spoof
   by an untrusted device or that would ensure that even if an attacker
   created such a packet, it would be dropped by the routers already
   deployed in the Internet today. One option of such marking described
   in this document is using a different protocol ID in the layer-2
   frames when sending IPv4 control packets among the routers. We call
   this "control IPv4 encapsulation". All Internet routers used today
   will drop these packets as unrecognized by default. This step makes
   sure that such a packet marking technique can be relied upon.

   The next step is a small modification to the router's local IP pro-
   cessing and encapsulation logic to allow only control-encapsulated
   IPv4 packets to be sent to the control plane along the normal path.
   Other packets are considered dangerous and are put on a heavily rate-
   limited queue. This ensures that outsider attacks do not exhaust
   resources used for communication with trusted devices.  Note that the
   encapsulation check has O(1) complexity, and can easily be performed
   at line rate even in legacy routers without major HW or SW modifica-
   tions. One of the many advantages of this approach is in the fact
   that no additional packet filtering at the customer or peering inter-
   faces is requires by the service provider, since user data packets
   always enter the network as dangerous.

   Note that the proposed mechanism does not require modification or
   affect existing Internet user data or network troubleshooting proto-
   cols--ICMP will still work they way it works today, so ping, tracer-
   oute, TCP path MTU discovery, remain functional. The reason for this
   is the fact that the proposal only helps routers quickly classify
   packets as trusted and untrusted, but does not require untrusted
   packets (e.g., ICMP) to be dropped. Of course, if a router already
   has a capability to identify ICMP packets and put them on a separate
   queue, the service provider may decide to configure the router to
   drop all untrusted packets except for ICMP.

   It should also be noted, that this proposal is not an attempt to pro-
   tect from compromised trusted routers or insider attacks, neither it
   is an attemp to substitute existing security mechanisms in routing
   protocols. Instead it helps to protect the routers from outsider
   (user-level) attacks, such as distributed DoS attacks based on infec-
   tion of untrusted devices (Internet-connected hosts) with computer
   viruses turning them into traffic generators targeted against Inter-
   net routers, which is considered to be a bigger threat today.  In the
   situation where a trusted router is compromised, the mechanism still

Zinin                                                           [Page 4]

INTERNET DRAFT            Router DoS protection            February 2003

   offers additional security by limiting the potential affect of the
   attack to the boundaries of the trust domain the compromised router
   participates in through the notion of interface groups. Routers
   implementing this mechanism and not participating in that domain will
   not be susceptible to the attack.

   Finally, the mechanism allows for gradual deployment across the
   Internet without a flag day and incremental security gain as it is
   deployed wider.

 2.2 Separating Data and Control Encapsulation

   As discussed before, the packet marking technique needs to have the
   property of default invalidity in order to make sure that no data
   flowing on the Internet today is considered trusted and is accepted
   into a service provider's network with such marking if an attacker
   tried to spoof a packet. Using techniques like DSCP-code marking or
   IP options does not satisfy this requirement, as it would call for
   filtering at every customer-facing router in the Internet to make
   sure that no user data packet is injected with this reserved DSCP
   value. This is the reason why the author has chosen to use a layer-2
   encapsulation technique to achieve this--frames carrying unknown pro-
   tocols are dropped by todays deployed routers.

   This document describes two possible methods for a different layer-2
   encapsulation--a separate protocol ID, and a link-local MPLS label.
   Each has its own advantages and disadvantages discussed below .

   Option 1: New Protocol ID

     As a protocol ID value is defined for IPv4 and IPv6 for each used
     media type today (such as Ether_type code), it would be possible to
     define IPv4-control and IPv6-control protocol IDs.

     The advantage of this method is an implicit 100% guarantee that if
     the protocol ID is selected from an unused space, the packets will
     be unrecognized. This approach also seems like the "clean" way of
     doing this.

     The disadvantage of this approach is that the control encapsulation
     protocol ID will need to be defined for each media type used today,
     which may take a while. Another disadvantage is that in case of an
     MPLS network, a control packet maybe put on an LSP together with
     data packets, so the receiving router wouldn't be able to tell the
     difference. Getting around this problem may require maintaining two
     sets of next-hops per route in the data path. See Option 3 below,

Zinin                                                           [Page 5]

INTERNET DRAFT            Router DoS protection            February 2003

   Option 2: Link-local MPLS Label

     This method is more of a hack and relies on the fact that MPLS
     encapsulation is either defined for or mapped to most of today's
     used media types. It should be possible to reserve a single label
     value (or 2 if a separate one for IPv6 is deemed necessary) from
     the "reserved" range (values 4-15 defined by [MPLS-STACK]), declare
     it to be link-local, disallow this value from being used for tran-
     sit MPLS LSPs, and use this as the control encapsulation. Note that
     since the label would only have significance on the local link, it
     can be reused on all links. Control messages used for signaling of
     transit label switched paths (LSPs) can be safely put on top of
     this label, as there are no order of origin dependencies. Routers
     that do not support MPLS would not need to have any MPLS code added
     and could just treat this as a special sequence of octects in the
     link frame that identifies control encapsulation.

     When a control packet for a multi-hop routing session (iBGP or OSPF
     virtual link) is put on an LSP, an extra label with the reserved
     value would be added on top of the label stack thus identifying the
     control packet.

     Because service providers generally do not support MPLS on their
     customer interfaces, and because the label value would be taken
     from the reserved space, it would be impossible for an Internet
     user to spoof a control packet using existing Internet infrastruc-

     The advantage of this approach is that only a single value for the
     label would need to be reserved.

     The disadvantages are that more modifications of the router
     microcode are necessary.

   Option 3: Combined

     It is possible to use the new protocol ID whenever a control packet
     is not MPLS-encapsulated, and use an extra reserved label whenever
     it is put on an LSP. See section XXX for more information on sup-
     port of MPLS networks.

   Control-plane software is then modified to make sure that all
   locally-originated packets that are relevant within the service
   provider's network only (such as routing protocols, MPLS signaling,
   telnet, ssh, SNMP, etc.) are control-encapsulated when the outbound
   interface is configured as such. Control packets that need to be
   received by the users (ICMP) are either encapsulated as before (data
   encapsulation) or also as control. In the latter case, they will be

Zinin                                                           [Page 6]

INTERNET DRAFT            Router DoS protection            February 2003

   data-encapsulated as soon as they leave the trust domain of the ser-
   vice provider.

 2.3 Interface Groups

   When deploying this mechanism, the service provider will need to
   identify a group of interfaces where the control encapsulation should
   or should not be used. There will most probably be a group of inter-
   faces used for the backbone connection, and another group used for
   customer connections and peering with other service providers.

   The described mechanism uses the notion of an "interface group".
   There is practically no complexity associated with an interface
   group--each interface has an interface-group attribute associated
   with it. Two interfaces are considered to be in one interface group
   if their interface-group attributes are equal. The service provider
   is expected to configure the interface group attributes of the inter-
   faces to match the trust communities, as in the following example.
   Backbone interfaces, interfaces to customer A, interfaces to customer
   B, interfaces to service provider X, and interfaces to service
   provider Y, would all be put in separate interface-groups: "back-
   bone", "cust-A", "cust-B", "peer-X", "peer-Y", correspondingly.

   As we will see further in the document, when a control-encapsulated
   packet is forwarded across an interface-group boundary, it become
   data-encapsulated (untrusted). This is to ensure that if, for exam-
   ple, two service providers are using control encapsulation for their
   eBGP session, or if an eBGP session between a service provider and a
   customer is control-encapsulated, forged packets originated by a
   potentially compromised BGP peer and destined inside of the service
   provider's network are not considered trusted beyond the border
   router. In other words, we trust control traffic from a customer or
   another service provider only as far as it needs to go and no fur-
   ther. Again, once a control-encapsulated packet crosses an interface-
   group boundary, its encapsulation is changed to data and it will be
   considered as untrusted by all other routers.

 2.4 Modified Local Processing and Packet Encapsulation Procedures

   The following new interface parameters used by the modified algo-
   rithms are introduced.

          the ID of the group the interface belongs to

          defines which encapsulation should be used on the interface to
          send control packets originated locally by the router or

Zinin                                                           [Page 7]

INTERNET DRAFT            Router DoS protection            February 2003

          received as control-encapsulated on another interface. Possi-
          ble values: Data, and Control. Default: Data.

          defines the type of encapsulation that needs to be used in
          order for the received packet to be allowed for local process-
          ing by the RP as trusted. Values: Data, Control, Both .
          Default: Data.

   The router's behavior is modified as follows.

     1.   A packet addressed to the router itself is considered trusted
          and is allowed to be locally processed (queued to the control
          card) if IpCtlRcvEncap of the receiving interface is set to
          Both, or matches the encapsulation that was used to send the
          received packet.  Otherwise, the packet is put on a "slow"
          queue (or dropped if the router has the capability to recog-
          nize ICMP packets and still allow them to be processed in a
          rate-limited fashion).

     2.   The router uses Control encapsulation for an outgoing packet
          if IpCtlSndEncap of the outbound interface is Control AND the

          a)   Has been locally originated by the router itself, OR

          b)   Has been received in Control encapsulation AND Interface-
               Group parameters of the inbound and outbound interfaces
               are the same (the packet is not leaving its trust

          Otherwise, (the packet is untrusted or is leaving its trust
          domain by crossing the interface-group boundary), Data encap-
          sulation is used.

 2.5 NOC Support and "Trusted" Interfaces

   Hosts on the NOC segments of the service provider's network are an
   example of trusted devices that are not routers. However, unlike
   routers, it is unrealistic to expect hosts within the NOC segment to
   exchange packets using Control encapsulation, as this would require
   modification to many operating systems. Another specific of a NOC
   segment is the fact that it majority of cases, it will need to be
   able to communicate with the rest of the service provider's network
   using both Data and Control encapsulated packets. The following is an
   explanation why.

   It is already a common practice to allow incoming telnet, ssh, and

Zinin                                                           [Page 8]

INTERNET DRAFT            Router DoS protection            February 2003

   snmp packets to routers only if they were originated within a NOC
   segment (this is usually done by configuring CLI and SNMP-specific
   filters by access control lists), however, the network administrators
   are rarely physically located on the NOC premises-many of them work
   from home and are often mobile. This is why management access to the
   routers usually requires establishing a secure shell (SSH) session to
   a server in the NOC, and then from there another SSH session to a
   router. Of course, the SSH server is usually behind a firewall (let's
   call it FW1). In our case, this would be a firewall that communicates
   Data packets with the rest of the service provider's network.

   Since all telnet, ssh, snmp, etc. packets going from the NOC segment
   to the routers in the network need to appear in Control encapsula-
   tion, regular data packets exchanged on the NOC segment at some point
   need to be sent out as Control encapsulated packets. This, of course,
   introduces a potential security threat (if the hosts on the NOC seg-
   ment were used to attack the routers, all forged packets would be
   considered by routers as trusted.) However, it is much less expensive
   for a service provider to protect its routers from its own NOC seg-
   ments by installing a firewall (let's call it FW2) that will make
   sure that only valid packets are sent out as control to the routers
   in the network.

   Note that FW1 and FW2 are only functionally separate, but may physi-
   cally be the same device.

   There are potentially two ways how NOC data packets can be injected
   as Control into the network: a) FW2's network-facing interface sup-
   ports Control encapsulation, and b) FW2 has no support of Control
   encapsulation, but the first-hop router it is connected to performs
   the "translation". The former case is the most secure, while the lat-
   ter is the most probable, at least in the beginning. Below is how the
   router performs the translation function.

   The notion of a "trusted interface" is defined by introducing the
   following parameter:

   IpTrustedInterface: When True, identifies a trusted interface. It is
   expected that only very few interfaces in the service provider's net-
   work will be configured as Trusted (for example, interfaces connect-
   ing a NOC segment to the rest of the network through a firewall.)
   Possible values: True, and False. Default: False.

   The router's behavior is further modified to accommodate the notion
   of trusted interface as follows:

     1.   A packet received on a Trusted interface in any encapsulation
          is treated as if it was received in Control encapsulation

Zinin                                                           [Page 9]

INTERNET DRAFT            Router DoS protection            February 2003

          (i.e., is allowed to be locally processed and is sent out
          using Control encapsulation as long as it stays within the
          same interface group).

     2.   All packets (trusted and untrusted) sent out of a Trusted
          interface are Data-encapsulated.

   DISCUSSION: we may want to allow only trusted packets to be sent on a
   Trusted interface towards NOC. This will make the job of FW2 much
   easier, but will cut off ICMP messages coming from outside the net-
   work or from a different trust domain if the service provider has

 2.6 ICMP, Ping, and Traceroute

   ICMP needs special attention, because its scope of validity is not so
   well contained as for routing and signaling protocols. Let's consider
   how this proposals handles ICMP by looking at the following generic
   combinations for ICMP messages:

     1.   Originated by and addressed to devices within the same trust
          domain, for example, an ICMP "Echo Request" message originated
          by a NOC host and received by a router. Same for an ICMP "Echo
          Reply". All ICMP messages will be considered trusted. No
          issues here.

     2.   Originated by a trusted device (router), addressed to an
          untrusted one (a ICMP "Destination Unreachable" to an Inter-
          net-connected host, for example). The router will inject the
          packet using Control encapsulation, however, as the packet
          leaves the service provider's network, it will be sent out
          using Data encapsulation (see step 2 in the modified router
          algorithm), as expected by the receiver.

     3.   Originated by an untrusted host, addressed to a service
          provider's router. The router will put the packet on the
          'untrusted' queue, and it will be processed.

     4.   Originated by and addressed to an untrusted host. The message
          will enter and leave the network as untrusted data without
          touching any router's control plane.

   Traceroute from the outside world does not present any problem,
   because Control-encapsulated ICMP messages sent back to the probing
   host will be automatically converted to Data as they leave the trust

   Traceroute within the network (e.g., from NOC or a router) is not a

Zinin                                                          [Page 10]

INTERNET DRAFT            Router DoS protection            February 2003

   problem because the messages are exchanged in Control encapsulation.
   If traceroute crosses multiple trust domains or goes outside the ser-
   vice provider's network, ICMP messages will come back as Data and
   will go through a FW to NOC or may be received through the slow queue
   by the router (if traceroute is originated by the router.)

 2.6 Routing Protocols

   One of the advantages of the described mechanism is that no modifica-
   tion of existing routing protocols is required. Routing protocols
   still work over IPv4, the only difference is actual layer-2 encapsu-
   lation of those packets, which is (in the simplest case) Control for
   all packets originated by a router.

   The routing paradigm remains the same--the messages are sent inbound
   across the same physical links as data packets. Control and data are
   only virtually separated, just enough to make a decision on whether a
   packet should be considered from a trusted source or not.

   Because the IS-IS routing protocol encapsulates its PDUs in L2
   frames, as opposed to IP packets, it is not susceptible to the out-
   sider attacks, and hence no modification to IS-IS encapsulation is
   required.  If IS-IS-in-IP is used, the routers need to make sure that
   the IP packets is Control-encapsulated. Note that the fact the IS-IS
   routing protocol is not susceptible to outsider attacks does not mean
   that ISP running IS-IS should not be worried about those attacks.
   There's a whole set of potential CPU-based attacks which an outsider
   could mount, and this set is constantly growing.

 2.7 Multicast

   There are two aspects of IP mutlicast we're interested in from the
   routing security point of view: routing protocols, and (S,G) state.

   From the routing protocols perspective, service provider's routers
   are protected by the presented mechanism as with unicast.

   The link between data and control plane required to maintain the
   (S,G) state is part of the multicast architecture and may be consid-
   ered by some as an issue (it is definitely safer to decouple control
   and data planes of the network as much as possible).  Presented secu-
   rity mechanism does not affect it in anyway. The service provider
   will have to make an informed decision whether to deploy multicast in
   its network or not keeping in mind the possibility of some router
   implementations not being able to keep up with large amounts of (S,G)

Zinin                                                          [Page 11]

INTERNET DRAFT            Router DoS protection            February 2003

 2.8 MPLS Networks

   When the mechanism is deployed in an MPLS network, it is possible for
   any IP packet (including a control one) that is sent over multiple
   hops to be put on an LSP due to an LSR using either LDP-derived FECs
   or IGP-shortcut FECs. Because layer-2 encapsulation is not preserved
   when an IP packet is put on an LSP, it will be impossible for the
   receiving router to tell the difference between data and control
   packets if just a different link-layer protocol ID was used to mark
   trusted packets.

   To solve this problem, the LSR putting the control packet on an LSP,
   adds an extra inner label with the reserved value described before to
   the label stack.

   If penultimate hop popping (PHP) is used in the network, the tail-end
   LSR may not even notice the fact that the packet has traveled on a
   LSP if the MPLS-label approach is used for encapsulation, because the
   LSR will receive the packet with only one--reserved--label.

   If the PHP mechanism is not used, the receiving LSR, after popping
   the outer label, will need to recognize the reserved value of the
   inner label and treat the packet as Control-encapsulated.

3 Deployment Considerations

   The following subsections discuss how the described mechanism would
   be deployed in a service provider's network. Note that we consider
   the final setup, after all transitional steps. The transition scenar-
   ios are described in a separate subsection

 3.1 Backbone-only Routers

   Routers where all interfaces are connected to internal links will
   most often have all of them configured to be in the same interface
   group.  It is possible of course, to have multiple Control trust
   domains within a single service provider's network if for example,
   BGP AS confederations are used. In this case, each member-AS would be
   a separate trust domain and some BGP speakers would have more than
   one interface group. One consideration related to running a network
   with multiple trust domains is the fact that control message that are
   not naturally scoped to a single trust domain (such as ICMP) will be
   encapsulated as Data once they leave the trust domain they have been
   originated in. This means that Control encapsulation-aware firewalls
   connecting the NOC segment need to also receive and process Data-
   encapsulated ICMP.

   Receiving and sending encapsulation of control packets would be set

Zinin                                                          [Page 12]

INTERNET DRAFT            Router DoS protection            February 2003

   to Control on all interfaces.

 3.2 Customer-facing Routers

   Customer facing routers will have more than one interface group.

   One group will be configured for all backbone links. In this group
   receive and send encapsulation will be configured as Control.

   For each customer, all interfaces providing connections to it will be
   configured as a separate interface group. The type of encapsulation
   is expected to be Data for a long time, before customer routers start
   supporting Control encapsulation. With Data encapsulation, the router
   is allowed to send Data-encapsulated control packets to the control
   plane CPU. Other packets, supposedly both valid data and potentially
   forged packets, are forwarded onwards to the network using Data
   encapsulation, so other routers in the network won't allow these
   packets to the control plane in case of an attack.

   When Control encapsulation is supported by the customer routers, the
   service provider will configure send and receive control packet
   encapsulation on those links to be Control. This will prevent DoS
   attacks on the customer-facing router on those links.

 3.3 Peer-facing Routers

   Peer-facing routers will be configured similar to the customer-facing
   routers. If the peering routers do not support Control encapsulation,
   the routers are configured to allow Data-encapsulated packets to be
   received by the control CPU. Potential attacks against the border
   router could be prevented by the BGP TTL hack (though implementing
   Control encapsulation seems easier.) service provider's internal
   routers will not be susceptible to the attacks originated in other
   service providers, because forged packets will be sent as Data and
   won't be allowed to the routers' control plane CPUs. When Control
   encapsulation is supported, the border router will be protected from
   the DoS attack on the links to those service providers supporting
   this technique.

   An important point to keep in mind here is the fact that trust
   domains of the service providers are not merged when they peer with
   each other. Links used to peer with other service providers are put
   in a separate interface group from the backbone interface group. This
   means that even if routers of another service provider are compro-
   mised and forged packets are sent as Control to us, they would first
   be translated to Data encapsulation by that service provider's border
   router, but even if they are not for some reason (or if the service
   provider's border router is compromised), our border router will

Zinin                                                          [Page 13]

INTERNET DRAFT            Router DoS protection            February 2003

   "translate" any forged control packets into Data as they cross the
   boundary between the peering and the backbone interface group.

 3.4 Internet Exchgange Points and Help from LAN Switches

   In the case where a LAN segment is used to connect devices under dif-
   ferent administrative control (as in the case of an Internet Exchange
   point, for example), it is possible that some connected devices may
   not be trusted enough for others to agree to receive control-encapsu-
   lated packets from them. In order to avoid line-rate source-based
   filtering, LAN switches may be equipped with a small bit of addi-
   tional functionality controlling whether inbound control-encapsulated
   packets are allowed on a specific port. Because this check is done on
   a per-port basis and the switch does not need to look further than
   the layer-2 frame, this check can easily be performed at the line
   rate without performance degradation. The LAN switch administrator
   would then have to ensure that control frames are allowed only from
   trusted devices.

   From the security perspective this essentially means that the service
   provider, normally only marginally trusting its IX peers, would need
   to trust the IX administrator's decision on whether the remote device
   is trusted or not, in other words, when a service provider agrees to
   accept control frames on an IX-connected interface, it essentially
   agrees to trust the security of that IX. However, though this means
   that IX-connected routers are less secure from each other, even if a
   trusted IX router is compromised, the effect of the attack is limited
   to the border router by the notion of the interface groups. Besides,
   IX-connected routers still have the same level of security against
   user-level attacks, which is thought of as a bigger threat than an
   attack from a compromised or untrusted IX-connected device.

 3.5 NOC

   As describe before, NOC segments can be connected to a service
   provider's network either through a Control-encapsulation-aware FW,
   or through a regular FW connected to a router implementing Trusted

 3.5 Transition Scenarios

   To be completed.

   Note that no flag day is required and gradual deployment gives incre-
   mental security increase.

4 Security Considerations

Zinin                                                          [Page 14]

INTERNET DRAFT            Router DoS protection            February 2003

   The described proposal does not claim to provide complete protection
   of routers against all types of attacks. Instead, it raises the bar
   by attempting to prevent attacks mounted by outsiders that have no
   access to the SP's network except for basic IP connectivity. These
   types of attacks are considered to be the immediate threat on the
   Internet routing system and the proposal attempts to protect against
   it without requiring expensive hardware upgrades. By virtually sepa-
   rating control and data packets, the level of security in IP networks
   is raised to the one normally found in ATM or Frame Relay networks,
   where routing and signaling are virtually out-of-band. This level of
   security is considered by many to be just enough to feel comfortable.

   Insider attacks, based on the physical access to the SP's equipment
   or on compromising a trusted device (such as a router or a NOC-
   attached host) are not prevented by this mechanism.

   The described proposal relies on the notion of a trust domain, which
   implies that if a router is configured to accept Control-encapsulated
   packets on an interface, the administrator administrator has full
   control of the devices attached to the segment and capable of sending
   Control-encapsulated packets (in reality, any connected device should
   be assumed to be capable of doing so), and those devices are autho-
   rized to send them. In other words, physical security needs to be
   insured by the SP. This practically means that no devices that with
   high probability can be compromised by an outside attacker (such as
   servers, or hosts) should be allowed on the segments used for router
   connections. Point-to-point links used between routers encourage this
   requirement by their very nature, while LAN segments require more
   attention to ensure no unauthorized devices have access to them.
   Fortunately, this is already the best current practice that the ser-
   vice providers follow. In the situations where a device connected to
   an otherwise trusted segment is considered to be highly susceptible
   to being compromised, some help from the LAN switch used to implement
   the segment is required. See Section XXX for more information.

   Finally, because the described mechanism does not prevent from
   insider attacks, it should not be considered as a substitute for
   existing or future authentication mechanisms in routing protocols or
   other security measures used in the service provider networks (e.g.,
   SSH).  Instead, they should be considered complimentary to each other
   and used together. In fact, the more elaborate and computationally
   expensive routing protocol-specific mechanisms become, the easier it
   will be for an outside attacker to bring a router to its knees, and
   the more important it will be to separate control and data encapsula-
   tion in the Internet.

4. Intellectual Property Considerations

Zinin                                                          [Page 15]

INTERNET DRAFT            Router DoS protection            February 2003

   The IETF takes no position regarding the validity or scope of any
   intellectual property or other rights that might be claimed to per-
   tain to the implementation or use of the technology described in this
   document or the extent to which any license under such rights might
   or might not be available; neither does it represent that it has made
   any effort to identify any such rights. Information on the IETF's
   procedures with respect to rights in standards-track and standards-
   related documentation can be found in BCP-11. Copies of claims of
   rights made available for publication and any assurances of licenses
   to be made available, or the result of an attempt made to obtain a
   general license or permission for the use of such proprietary rights
   by implementors or users of this specification can be obtained from
   the IETF Secretariat.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights which may cover technology that may be required to practice
   this standard. Please address the information to the IETF Executive

   The IETF has been notified of intellectual property rights claimed in
   regard to some or all of the specification contained in this docu-
   ment.  For more information consult the online list of claimed

5. Acknowledgements

   Thanks to Ben Crosby, Steve Buchko, Randy Bush, John Heasley, Radia
   Perlman, and Tony Li for an early review of this work.

6. References

   [OSPF] J. Moy. OSPF version 2. Technical Report RFC 2328, Internet
          Engineering Task Force, 1998.

   [TCP-MD5] A. Hefferman. Protection of BGP Sessions via the TCP MD5
          Signature Option. RFC 2385. 1998.

   [MPLS-STACK] RFC 3032. MPLS Label Stack Encoding.

7. Authors' Addresses

   Alex Zinin

Zinin                                                          [Page 16]

INTERNET DRAFT            Router DoS protection            February 2003

Full Copyright Statement

   Copyright (C) The Internet Society (date). 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 implmentation 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 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

Zinin                                                          [Page 17]