Network Working Group Pedro Marques
INTERNET DRAFT Nischal Sheth
Expiration Date: December 2003 Juniper Networks, Inc
Robert Raszuk
Cisco Systems, Inc
Jared Mauch
NTT/Verio
Danny McPherson
Arbor Networks
June 2003
Dissemination of flow specification rules
draft-marques-idr-flow-spec-00.txt
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 specifies the procedures necessary for the distribution
of flow specification rules both intra an inter-as. Additionally it
defines its application for the purpose of traffic filtering in order
to mitigate (distributed) denial of service attacks.
The information is carried via the Border Gateway Protocol (BGP),
thereby reusing protocol algorithms, operational experience and
administrative processes such as inter-provider peering agreements.
draft-marques-idr-flow-spec-00.txt [Page 1]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
1. Introduction
For several applications, it may be necessary to exchange control
information pertaining to aggregated traffic flow definitions which
cannot be expressed using destination address prefixes only.
An aggregated traffic flow is considered to be an n-tuple consisting
on several matching criteria such as source and destination address
prefixes, IP protocol and transport protocol port numbers.
The intention of this document is to define a general procedure to
encode such flow specification rules as a BGP NLRI which can be
reused for several different control applications. Additionally, we
define the required mechanisms to utilize this definition to the
problem of immediate concern to the authors: intra and inter provider
distribution of traffic filtering rules to filter (Distributed)
Denial of Service (DoS) attacks.
The choice of BGP as the carrier of this control information is
justified by the fact that the key issues in terms of complexity are
problems which are common to unicast route distribution and have
already been solved in the current environment.
From an algorithmic perspective, the main problem that presents
itself is the distributed loop-free distribution of <key, attribute>
pairs. The key, in this particular instance, being a flow
specification.
From an operational perspective, the utilization of BGP as the
carrier for this information, allows a network service provider to
reuse both internal route distribution infrastructure (e.g.: route
reflector or confederation design) and existing external
relationships (e.g.: inter-domain BGP sessions to a customer
network).
While it is certainly possible to address this problem using other
mechanisms the authors believe that this solution offers the
substantial advantage of being an incremental addition to deployed
mechanisms.
draft-marques-idr-flow-spec-00.txt [Page 2]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
2. Flow specifications
A flow specification is an n-tuple consisting on several matching
criteria that can be applied to IP traffic. A given IP packet is said
to match the defined flow if it matches all the specified criteria.
A given flow may be associated with a set of attributes, depending on
the particular application, such attributes may or may not include
reachability information (i.e. NEXT_HOP). Well-known or AS-specific
community attributes can be used to encode a set of predeterminate
actions.
A particular application is identified by a specific (AFI, SAFI) pair
and corresponds to a distinct set of RIBs. Those RIBs should be
treated independently from each other in order to assure non-
interference between distinct applications.
BGP itself treats the NLRI as an opaque key to an entry in its
databases. Entries that are placed in the Loc-RIB are then associated
with a given set of semantics which is application dependent. This is
consistent with existing BGP applications. For instance IP unicast
routing (AFI=1, SAFI=1) and IP multicast reverse-path information
(AFI=1, SAFI=2) are handled by BGP without any particular semantics
being associated with them until installed in the Loc-RIB.
Standard BGP policy mechanisms, such as UPDATE filtering by NLRI
prefix and community matching, SHOULD apply to the newly defined
NLRI-type. Network operators can also control propagation of such
routing updates by enabling or disabling the exchange of a particular
(AFI, SAFI) pair on a given BGP peering session.
3. Dissemination of Information
We define a "Flow Specification" NLRI type that may include several
components such as destination prefix, source prefix, protocol,
ports, etc. This NLRI is treated as an opaque bit string prefix by
BGP. Each bit string identifies a key to a database entry which a set
of attributes can be associated with.
This NLRI information is encoded using MP_REACH_NLRI and
MP_UNREACH_NLRI attributes as defined in [BGP-MP]. Whenever the
corresponding application does not require Next Hop information, this
shall be encoded as a 0 octet length Next Hop in the MP_REACH_NLRI
attribute and ignored on receipt.
The NLRI field of the MP_REACH_NLRI and MP_UNREACH_NLRI is encoded as
a two byte NLRI length value in octets followed by a variable length
draft-marques-idr-flow-spec-00.txt [Page 3]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
NLRI value.
+------------------------------+
| NLRI length (2 octets) |
+------------------------------+
| NLRI value (variable) |
+------------------------------+
The Flow Specification NLRI-type consists of several optional
subcomponents. A specific packet is considered to match the flow
specification when it matches the intersection (AND) of all the
components present in the specification.
The following component types are defined:
Type 1 - Destination Prefix
Encoding: <type (1 octet), prefix length (1 octet), prefix>
Defines the destination prefix to match. Prefixes are encoded
as in BGP UPDATE messages, a length in bits is followed by
enough octets to contain the prefix information.
Type 2 - Source Prefix
Encoding: <type (1 octet), prefix-length (1 octet), prefix>
Defines the source prefix to match.
Type 3 - Protocol
Encoding: <type (1 octet), IP protocol (1 octet)>
Defines the IP protocol value to match.
Type 4 - Port
Encoding: <type (1 octet), TCP/UDP port (2 octets)>
Defines the TCP/UDP port to match as source OR destination.
Type 5 - Destination port
Encoding: <type (1 octet), TCP/UDP port (2 octets)>
Defines the destination TCP/UDP port to match.
Type 6 - Source port
draft-marques-idr-flow-spec-00.txt [Page 4]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
Encoding: <type (1 octet), TCP/UDP port (2 octets)>
Defines the source TCP/UDP port to match.
Type 7 - ICMP type
Encoding: <type (1 octet), icmp type(1 octet)>
Defines the source icmp type to match.
Type 8 - Packet length
Encoding: <type (1 octet), IP packet length (2 octets)>
Match on the total IP packet length (excluding L2 but including
IP header).
Flow specification components must follow strict type ordering. A
given component type may or may not be present in the specification,
but if present it MUST precede any component of higher numeric type
value.
If a given component type within a prefix in unknown, the prefix in
question cannot be used for traffic filtering purposes by the
receiver. Since a Flow Specification as the semantics of a logical
AND of all components, if a component is FALSE by definition it
cannot be applied. However for the purposes of BGP route propagation
this prefix should still be transmitted since BGP route distribution
is independent on NLRI semantics.
Flow specification components are to be interpreted as a bit match at
a given packet offset. When more than one component in a flow
specification tests the same packet offset the behavior is
undetermined.
The <type, value> encoding is chosen in order to account for future
extensibility.
An example of a Flow Specification encoding for: "all packets to
10.0.1/24 and TCP port 25".
destination proto port
+-------------+-----+--------+
01 18 0a 01 01 03 06 04 00 19
This constitutes a NLRI with an NLRI length of 10 octets.
Implementations wishing to exchange flow specification rules MUST use
draft-marques-idr-flow-spec-00.txt [Page 5]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
BGP's Capability Advertisement facility to exchange the Multiprotocol
Extension Capability Code (Code 1) as defined in [BGP-MP]. The (AFI,
SAFI) pair carried in the Multiprotocol Extension capability MUST be
the same as the one used to identify a particular application that
uses this NLRI-type.
4. Traffic filtering
Traffic filtering policies where traditionally considered to be
relatively static. The popularity of traffic-based denial of service
(DoS) attacks, which often requires the network operator to be able
to use traffic filters for detection and mitigation, brings with it
requirements that are not fully satisfied by existing tools.
Several techniques are currently used to control traffic filtering of
DoS attacks. Among those, one of the most common is to inject
unicast route advertisements corresponding to a destination prefix
being attacked. One variant of this technique marks such route
advertisements with a community that gets translated into a discard
next-hop by the receiving router. Other variants, attract traffic to
a particular node that serves as a deterministic drop point.
Using unicast routing advertisements to distribute traffic filtering
information has the advantage of using the existing infrastructure
and inter-as communication channels. This can allow, for instance,
for a service provider to accept filtering requests from customers
for address space they own.
There are several drawbacks, however. An issue that is immediately
apparent is the granularity of filtering control: only destination
prefixes may be specified. Another area of concern is the fact that
filtering information is intermingled with routing information.
The mechanism defined in this document is designed to address these
limitations. We use the flow specification NLRI defined above to
convey information about traffic filtering rules for traffic that
should be discarded.
This mechanism is designed to, primarily, allow an upstream
autonomous system to perform inbound filtering, in their ingress
routers of traffic that a given downstream AS wishes to drop.
In order to achieve that goal, we define an application specific NLRI
identifier (AFI=1, SAFI=TBD) along with specific sematic rules. BGP
routing updates containing this identifier use the flow specification
NLRI encoding to convey particular aggregated flows that require
special treatment.
draft-marques-idr-flow-spec-00.txt [Page 6]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
5. Validation procedure
Flow specifications received from a BGP peer and which are accepted
in the respective Adj-RIB-In are used as input to the route selection
process. Although the forwarding attributes of two routes for the
same Flow Specification prefix may be the same, BGP is still required
to perform its path selection algorithm in order to select the
correct set of attributes to advertise.
The first step of the BGP Route Selection procedure [BGP-BASE]
(section 9.1.2) is to exclude from the selection procedure routes
that are considered non-feasible. In the context of IP routing
information this step is used to validate that the NEXT_HOP attribute
of a given route is resolvable.
The concept can be extended, in the case of Flow Specification NLRI,
to allow other validation procedures.
A flow specification NLRI SHOULD be validated such that it is
considered unfeasible if it contains an non-empty AS_PATH and that
AS_PATH does not match the AS_PATH of the best match unicast route
that includes the specified destination address prefix.
The underlying concept is that the neighboring AS that advertises the
best unicast route for a destination is allowed to advertise flow
spec information that conveys a less or equally specific destination
prefix.
The neighboring AS is the immediate destination of the traffic
described by the Flow Specification. If it requests these flows to be
dropped that request can be honored without concern that it
represents a denial of service in itself. Supposedly, the traffic is
being dropped by the downstream autonomous-system and there is no
added value in carrying the traffic to it.
BGP implementations MUST also enforce that the AS_PATH attribute of a
route received via eBGP contains the neighboring AS in the left-most
position of the AS_PATH attribute. While this rule is optional in the
BGP specification, it becomes necessary to enforce it for security
reasons.
draft-marques-idr-flow-spec-00.txt [Page 7]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
6. Traffic Filtering Actions
The default action for a traffic filtering flow specification is to
discard IP traffic that matches that particular rule.
In some situations, a network operator may choose, to monitor a
particular aggregated flow while still forwarding the affected
packets. This action is specified by tagging the respective flow
specifications with the well-known community of value TBD.
Other actions such as rate-limiting or policy-based routing may be
employed as well, although further discussions of these actions is
considered beyond the scope of this document.
7. Monitoring
Traffic filtering applications require monitoring and traffic
statistics facilities. While this is an implementation specific
choice, implementations SHOULD provide:
- A mechanism to log the packet header of filtered traffic,
- A mechanism to count the number of drops matching a given Flow
Specification rule.
8. Security considerations
Inter-provider routing is based on a web of trust. Neighboring
autonomous-systems are trusted to advertise valid reachability
information. If this trust model is violated, a neighboring
autonomous system may cause a denial of service attack by advertising
reachability information for a given prefix for which it does not
provide service.
As long as traffic filtering rules are restricted to match the
corresponding unicast routing paths for the relevant prefixes, the
security characteristics of this proposal are equivalent to the
existing security properties of BGP unicast routing.
Where it not the case, this would open the door to further denial of
service attacks.
draft-marques-idr-flow-spec-00.txt [Page 8]
Internet Draft draft-marques-idr-flow-spec-00.txt June 2003
9. Acknowledgments
The authors would like to thank Yakov Rekhter, Dennis Ferguson and
Chris Morrow for their comments.
10. References
[BGP-BASE] Y. Rekhter, T. Li, S. Hares, "A Border Gateway Protocol 4
(BGP-4)", draft-ietf-idr-bgp4-20.txt, 03/03
[BGP-MP] T. Bates, R. Chandra, D. Katz, Y. Rekhter, "Multiprotocol
Extensions for BGP-4", RFC2858.
11. Authors' Addresses
Pedro Marques
Juniper Networks
1194 N. Mathilda Ave.
Sunnyvale, CA 94089
E-mail: roque@juniper.net
Nischal Sheth
Juniper Networks
1194 N. Mathilda Ave.
Sunnyvale, CA 94089
E-mail: roque@juniper.net
Robert Raszuk
Cisco Systems, Inc.
Al. Jerozolimskie 146C
02-305 Warsaw, Poland
Email: rraszuk@cisco.com
Jared Mauch
NTT/VERIO
8285 Reese Lane
Ann Arbor, MI, 48103-9753
Email: jmauch@verio.net | jared@puck.nether.net
Danny McPherson
Arbor Networks
Email: danny@arbor.net
draft-marques-idr-flow-spec-00.txt [Page 9]