Internet Engineering Task Force Ken Carlberg INTERNET DRAFT G11 June 9, 2004 A Framework for Supporting ETS Within a Single Administrative Domain Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026 [1]. 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. For potential updates to the above required-text see: http://www.ietf.org/ietf/1id-guidelines.txt Abstract This document presents a framework discussing the role of various protocols and mechanisms that could be considered candidates for supporting ETS within a single administrative domain. Comments about their potential usage as well as their current deployment are provided to the reader. Specific solutions are not presented. 1. Introduction This document presents a framework for supporting Emergency Telecommunications Service (ETS) within the scope of a single administrative domain. This narrow scope provides a reference point for considering protocols that could be deployed to support ETS. [2] is a complimentary effort that articulates requirements for a single administrative domain. We use this other effort as both a starting point and guide for this document. A different example of a framework document for ETS is [3], which Carlberg Expires December 9, 2004 [Page 1] Internet Draft ETS Single Domain Framework June 9, 2004 focused on support for ETS within IP telephony. In this case, the focal point was a particular application whose flows could span multiple autonomous domains. Even though this document uses a somewhat more constrained perspective than [3], we can still expect some measure of overlap in the areas that are discussed. 1.1 Differences between Single and Inter-domain The progression of our work in the following sections is helped by stating some key differences between the single and inter-domain cases, which are articulated in the following: a) Different policies might be implemented in different administrative domains. b) There is an absence of any practical method for authenticating all of the network layer packets that have labels indicating a preference or importance at ingress nodes to other administrative domains (e.g. on the uplink into an ISP). c) Given item (b) above, all current inter-domain QoS mechanisms at the network level generally create easily exploited and significantly painful DoS/DDoS attack vectors on the network. d) A single administrative domain can deploy various mechanisms (e.g. Access Control Lists) into each and every edge device (e.g. ethernet switch or router) to ensure that only authorized end-users (or layer 2 interfaces) are able to emit frames/packets with non-default QoS labels into the network. This is not feasable in the inter-domain case because the inter-domain link contains aggregated flows. In addition, the dissemination of Access Control Lists at the network level is not scalable in the inter-domain case. e) A single domain can deploy mechanisms into the edge devices to enforce its domain-wide policies -- without having to trust any 3rd party to configure things correctly. This is not possible in the inter-domain case. While the above is not an all-inclusive set of differences, it does provide some rationale why one may wish to focus efforts in the more constrained scenario of a single administrative domain. 2. Common Practice: Provisioning The IEPREP working group, and mailing list, has had extentive discussions about over-provisioning. Many of these exchanges have debated the need for QoS mechanisms versus over-provisioning of Carlberg Expires December 9, 2004 [Page 2] Internet Draft ETS Single Domain Framework June 9, 2004 links. In reality, nearly all IP network links are over-provisioned with excess capacity for the average load. The 'shared' resource model together with TCP's congestion avoidance algorithms helps compensate for those cases where spikes or bursts of traffic are experienced by the network. The thrust of the debate within the IEPREP working group is whether links should be over provisioned to such a degree that spikes in load can still be supported with no loss due to congestion. Advocates of this position point to many ISPs in the U.S. that take this approach instead of using QoS mechanisms to honor agreements with their peers or customers. These advocates point to cost effectiveness in comparison to complexity and security issues associated with other approaches. This document does not advocate one of these positions over the other. The author does advocate that network administrators/operators should perform a cost analysis between over provisioning for spikes versus QoS mechanisms. This analysis, in addition to examining policies and requirements of the administrative domain, should be the key to deciding how (or if) ETS should be supported within the stub domain. If the decision is to rely on over provisioning, then some of the following sections will have little to no bearing on how ETS is supported within a stub domain. The exception would be labeling mechanisms used to convey information to other communication architectures (e.g., SIP-to-SS7/ISUP gateways). 3. Objective The primary objective is to provide a target measure of service within a stub domain for flows that have been labeled for ETS. This level may be better than best effort, or it may be the best available service that the network (or parts thereof) can offer. [2] presents a set of requirements in trying to achieve this objective. This framework document uses [2] as a reference point in discussing existing areas of engineering work or protocols that can play a role in supporting ETS within a domain. Discussion of these areas and protocols are not to be confused with expectations that they exist within a given domain. Rather, the subjects discussed in Section 4 below are ones that are recognized as candidates that can exist and could be used to facilitate ETS users or data flows. 3.1 Scenarios Carlberg Expires December 9, 2004 [Page 3] Internet Draft ETS Single Domain Framework June 9, 2004 One of the topics of discussion that arises on the IEPREP mailing list, and the working group meetings, is the operating environment of the ETS user. Many variations can be dreamed of with respect to underlying network technologies and applications. Instead of getting lost in hundreds of potential scenarios, we attempt to abstract the limit the scenarios into two simple case examples. (a) A user in the HOME network attempts to use or leverage any ETS capability within the stub domain. (b) A user visits a FOREIGN network and attempts to use or leverage any ETS capability within the stub domain. We borrow the terms "home" and "foreign" network from that used in Mobile IP [4]. Case (a) is considered the normal and vastly most prevalent scenario in today's Internet. Case (b) above may simply be supported by the Dynamic Host Configuration Protocol (DHCP) [5], or a static set of addresses, that are assigned to 'visitors' of the network. This effort is predominantly operational in nature and heavily reliant on the management and security policies of that stub network. A more ambitious way of supporting the mobile user is through the use of the Mobile IP (MIP) protocol. In this case and at the IP level, foreign networks introduce the concept of triangle routing and the potential for multiple access points and service context within a subnetwork. In addition, policy plays a criticial role in dictating the measure of available services to the mobile user. The beaconing capability of MIP allows it to offer a measure of application transparent mobility as a mobile host (MH) moves from one subnetwork to another. However, this feature may not be available in most domains. In addition, its management requirements may discourage its widespread deployment and use. Hence, users should probably not rely on its existance, but rather may want to expect a more simpler approach based on DHCP as described above. The subject of mobile IP is discussed below in Section 4. 4. Topic Areas The topic areas presented below are not presented in any particular order or along any specific layering model. They represent capabilities that may be found within an administrative domain. Many are topics of on-going work within the IETF. It must be stressed that readers of this document should not expect any of the following to exist within a for ETS users. In many cases, while some of the following areas have been standardized for several Carlberg Expires December 9, 2004 [Page 4] Internet Draft ETS Single Domain Framework June 9, 2004 years, many have seen very limited deployment. 4.1 MPLS Multi-Protocol Label Switching (MPLS) is generally the first protocol that comes to mind when the subject of traffic engineering is brought up. MPLS is an intra-domain routing protocol that produces Labeled Switched Paths (LSP) through a network [6]. When traffic reaches the ingress boundary of an MPLS domain (which may or may not be congruent with an administrative domain), the packets are classified, labeled, scheduled, and forwarded along an LSP. [7] is an RFC describing how MPLS can support Differentiated Services. The RFC discusses the use of the 3 bit EXP (experimental) field to convey the Per Hop Behavior (PHB) to be applied to the packet. As we shall see in later subsections, this three bit field can be mapped to fields in several other protocols. The inherent feature of classification, scheduling, and labeling are viewed as symbiotic and therefore many times it is integrated with other protocols and architectures. Examples of this include RSVP and Differentiated Services. Below, we discuss several instances where a given protocol specification or mechanism has been known to be complimented with MPLS. This includes the potential labels that may be associated with ETS. However, we stress that MPLS is only one of several approaches to support traffic engineering. In addition, the complexity of the MPLS protocol and architecture may make it suited for only large domains. 4.2 RSVP The original design of RSVP, together with the Integrated Services model, was one of an end-to-end capability that spanned networks and administrative domains [8]. Currently, RSVP has not been widely deployed, and the limited deployment so far has been mostly constrained to boundaries within a domain. Early deployments of RSVP ran into unanticipated scaling issues; it is not entirely clear how scalable an RSVP approach would be across the Internet. Also, currently many network products do not support RSVP for anything beyond simple MPLS signalling. [9] is one example of how RSVP has evolved to compliment efforts that are scoped to operate within a domain. In this case, extentions to RSVP are defined that allow it to establish intra-domain Labled Switched Paths (LSP) in Multi-Protocol Labeled Switching (MPLS). [10] specifies extentions to RSVP so that it can support generic policy based admission control. This standard goes beyond the Carlberg Expires December 9, 2004 [Page 5] Internet Draft ETS Single Domain Framework June 9, 2004 support of the POLICY_DATA object stipulated in [9], by defining the means of control and enforcement of access and usage policies. While the standard does not advocate a particular policy architecture, the IETF has defined one that can compliment [10] -- we expand on this in subsection 4.3 below. 4.2.1 Relation to ETS The ability to reserve resources correlates to an ability to provide preferential service for specifically classified traffic -- the classification being a tuple of 1 or more fields. In cases where a tuple includes a label that has been defined for ETS usage, the reservation helps ensure that an emergency related flow will be forwarded towards its destination. Within the scope of this document, this means that RSVP would be used to facilitate the forwarding of traffic within a stub domain. We note that this places an importance on defining a label and an associated field that can be set and/or examined by RSVP capable nodes. Another important observation is that major vendor routers currently constrain their examination of fields for classification to the network and transport layers. This means that application layer labels will mostly likely be ignored by routers/switches. Thus, endpoints (which may be a server or proxy) that intend to add RSVP support for ETS should map application layer ETS labels to labels at the network or transport layer. 4.3 Policy The Common Open Policy Service (COPS) protocol [11] was defined to provide policy control over QoS signaling protocols, such as RSVP. COPS is based on a query/response model in which Policy Enforcement Points (PEPs) interact with Policy Decision Points (i.e., policy servers) to exchange policy information. COPs provides application level security and can operate over IPSEC or TLS. COPS is also stateful protocol that also supports a push model. This means that servers can download new policies, or alter existing ones to known clients. [12] articulates the usage of COPS with RSVP. This document specifies COPS client types, context objects, and decision objects. Thus, when an RSVP reservation is received by a PEP, the PEP decides whether to accept or reject it based on policy. This policy information can be stored a priori to the reception of the RSVP PATH message, or it can be retrieved in an on-demand basis. A similar course of action could be applied in cases where ETS labeled control Carlberg Expires December 9, 2004 [Page 6] Internet Draft ETS Single Domain Framework June 9, 2004 flows are received by the PEP. This of course would require an associated (and new) set of documents that first articulates types of ETS signaling and then specifies its usage with COPS. A complimentary document to the COPS protocols is [13], which describes the use of COPS for policy provisioning. As a side note, the current lack in deployment of RSVP in network products has also played at least an indirect role in the subsequent lack of implementations & deployment of COPS. [14] is an additional source for recent thoughts on this subject. 4.4 Subnetwork Technologies This is a generalization of work that is considered "under" IP and for the most part outside of the IETF standards body. We discuss some specific topics here because there is a relationship between them and IP in the sense that each physical network interacts at its edge with IP. 4.4.1 802.1 The IEEE 802.1q standard defined a tag appended to a Media Access Controller (MAC) frame for support of layer 2 Virtual Local Area Networks (VLAN). This tag has two parts: a VLAN identifier (12 bits) and a Prioritiation field of three bits. A subsequent standard, IEEE 802.1p, later incorporated into a revision of IEEE 802.1d, defined the Prioritization field of this new tag [15]. It consists of eight levels of priority, with the highest priority being a value of 7. Vendors may choose a queue per priority codepoint, or aggregate several codepoints to a single queue. The three bit Prioritization field can be easily mapped to the old ToS field of the upper layer IP header. In turn, these bits can also be mapped to a subset of differentiated code points. Bits in the IP header that could be used to support ETS (e.g., specific Diff-Serv code points) can in turn be mapped to the Prioritization bits of 802.1p. This mapping could be accomplished in a one-to-one manner between the 802.1p field and the IP ToS bits, or in an aggregate manner if one considers the entire Diff-Serv field in the IP header. In either case, because of the scarcity of bits, ETS users should expect that their traffic will be combined or aggregated with the same level of priority as some other types of "important" traffic. In other words, given the existing 3 bit Priority Field for 802.1p, there will not be an exclusive bit reserved for ETS traffic. Certain vendors are currently providing mappings between 802.1p field and the ToS bits. This is in addition to integrating the signaling Carlberg Expires December 9, 2004 [Page 7] Internet Draft ETS Single Domain Framework June 9, 2004 of RSVP with the low level inband signaling offered in the Priority field of 802.1p. It is important to note that the 802.1p standard does not specify the correlation of a layer 2 codepoint to a physical network bandwidth reservation. Instead, this standard provides what has been termed as "best effort QoS". The value of the 802.1p Priority code points is realized at the edges: either as the MAC payload is passed to upper layers (like IP), or bridged to other physical networks like Frame Relay. Either of these actions help provide an intra-domain wide propagation of a labeled flow for both layer 2 and layer 3 flows. 4.4.2 Cable Networks The Data Over Cable Service Interface Specification (DOCSIS) is a standard used to facilitate the communication and interaction of the cable subnetwork with upper layer IP networks [16]. Cable subnetworks tend to be asynchronous in terms of data load capacity: typically, 27M downstream, and anywhere from 320kb to 10M upstream (i.e., in the direction of the user towards the Internet). The evolution of the DOCSIS specification, from 1.0 to 1.1, brought about changes to support a service other than best effort. One of the changes was indirectly added when the 802.1D protocol added the Priority field, which was incorporated within the DOCSIS 1.1 specification. Another change was the ability to perform packet fragmentation of large packets so that Priority marked packets (i.e., packets marked with non-best effort labels) can be multiplexed inbetween the fragmented larger packet. Its important to note that the DOCSIS specifications do not specify how vendors implement classification, policing, and scheduling of traffic. Hence, operators must rely on mechanisms in Cable Modem Termination Systems (CMTS) and edge routers to leverage indirectly or directly the added specifications of DOCSIS 1.1. As in the case of 802.1p, ETS labeled traffic would most likely be aggregated with other types of traffic, which implies that an exclusive bit (or set of bits) will not be reserved for ETS users. Policies and other managed configurations will determine the form of the service experienced by ETS labeled traffic. Traffic engineering and management of ETS labeled flows, including its classification and scheduling at the edges of the DOCSIS cloud, could be accomplished in several ways. A simple schema could be based on non-FIFO queuing mechanisms like class based queuing, weighted fair queuing (or combinations and derivations thereof). The addition of active queue management like Random Early Detection could provide simple mechanisms for dealing with bursty traffic Carlberg Expires December 9, 2004 [Page 8] Internet Draft ETS Single Domain Framework June 9, 2004 contributing to congestion. A more elaborate scheme for traffic engineering would include the use of MPLS. However, the complexity of MPLS should be taken into consideration before its deployment in networks. 4.5 Multicast Network layer multicast has existed for quite a few years. Efforts such as the Mbone have provided a form of tunneled multicast that spans domains, but the routing hierarchy of the Mbone can be considered flat and non-congruent with unicast routing. Efforts like the Multicast Source Discovery Protocol [17] together with the Protocol Independent Multicast Sparse Mode (PIM-SM) have been used by a small subset of Internet Service Providers to provide form of inter-domain multicast [18]. However, network layer multicast has for the most part not been accepted as a common production level service by a vast majority of ISPs. In contrast, intra-domain multicast in stub domains has gained more acceptance as an additional network service. This support is further enhanced by corresponding support by physical networks. 4.5.1 IP Layer The value of IP multicast is its efficient use of resources in sending the same datagram to multiple receivers. An extensive discussion on the strengths and concerns about multicast is outside the scope of this document. However, one can argue that multicast can very naturally compliment the push-to-talk feature of land mobile radio networks (LMR). Push-to-talk is a form of group communication where every user in the "talk group" can participate in the same conversation. LMR is the type of network used by First Responders (e.g., police, fireman, and medical personnel) that are involved in emergencies. Currently, certain vendors and providers are offering push-to-talk service to the general public in addition to First Responders. Some of these systems are operated over IP networks, or are interfaced with IP networks to extend the set of users that can communicate with each other. We can consider at least a subset of these systems as either closed IP networks, or stub domains since they do not act as transits to other parts of the Internet. The potential integration of LMR talk groups with IP multicast is an open issue. LMR talk groups are established in a static manner with man-in-the-loop participation in their establishement and teardown. The seamless integration of these talk groups with multicast group addresses is a feature that has not been discussed in open forums. Carlberg Expires December 9, 2004 [Page 9] Internet Draft ETS Single Domain Framework June 9, 2004 4.5.2 802.1d The IEEE 802.1d standard specifies fields and capabilities for a number of features. In subsection 4.3.2 above, we discussed its use for defining a Prioritization field. The 802.1d standard also covers the topic of filtering MAC layer multicast frames. One of the concerns about multicast are broadcast storms that can arise and generate a denial of service against other users/nodes. Some administrators purposely filter out multicast frames in cases where the subnetwork resource is relatively small (e.g., 802.11 networks). Operational considerations with respect to ETS may wish to consider doing this in an as-needed basis based on the conditions of the network against the perceived need for multicast. In cases where filtering out multicast can be activated dynamically, COPS may be a good means of providing consistent domain-wide policy control. 4.6 Discovery If a service is being offered to explicitly support ETS, then it would seem reasonable that discovery of the service may be of benefit. For example, if a domain has a subset of servers that recognize ETS labeled traffic, then dynamic discovery of where these servers are (or even if they exist) would be more benefitial compared to relying on statically configured information. The Service Location Protocol (SLP) [19] is designed to provide information about the existance, location, and configuration of networked services. In many cases, the name of the host supporting the desired service is needed to be known a priori in order for users to access it. SLP eliminates this requirement by using a descriptive model that identifies the service. Based on this description, SLP then resolves the network address of the service and returns this information to the requester. An interesting design element of SLP is that it assumes that the protocol is run over a collection of nodes that are under the control of a single administrative authority. This model follows the scope of this framework document. Anycasting [20] is another means of discovering nodes that support a given service. Interdomain anycast addresses, propagated by BGP, have been used by one of the root servers for a while. [21] covers the topic of anycast addresses for IPv6. Unlike SLP, users/applications must know the anycast address associated with the target service. The tradeoffs between this approach and SLP is outside the scope of this framework document. 4.7 Mobility Carlberg Expires December 9, 2004 [Page 10] Internet Draft ETS Single Domain Framework June 9, 2004 The mobile user extends the scenario of how an ETS user operates within a domain. While the ownership of the mobile host may be different from other nodes in the same domain, the management of that node in terms of policies and administration is still defined by the foreign network (i.e., stub domain) that it is attached to. 4.7.1 Mobile IP Currently within the IETF, the subject of mobility is addressed in several ways. The oldest and most mature area involves mobile hosts and its support based on the Mobile IP protocol [RFC3344]. In this case, mobility is kept transparent from the upper layers and its support is focused at the network layer. The Mobile IP protocol (MIP) and architecture addresses the fundamental characteristics of a ETS user migrating to a foreign network and attempting to contact other users. One can also make an arguement that the percieved needs of an ETS user, e.g., labeling traffic to distinguish it from other flows can also be acheived independent of the MIP. A potential exception to this position is the "busy" bit that can be set during the initial registration of the Mobile Host (MH) to the Foreign Network. If the bit is tied to a maximum number of simultaneous number of MHs, then it may be of interest to specify an extension that distinguishes an ETS capable MH from other MHs. Local policy would determine the course of action to be taken. 4.7.2 Other Areas of Mobility As of the publication of this document, there are other working groups within the IETF that are involved in mobility. The Mobile Ad-Hoc Networking (MANET) working group has focused on the case in which all nodes, routers and hosts, can move in relation to each other. The output of this group has been in the form of experimental protocols, and so the subject area may be considered too immature in considering how it and the various protocols can play a role in supporting ETS. The Network Mobile (NEMO) working group has just recently been formed to address the issues that arise when entire networks move in relation to each other. This effort can currently be considered too immature for supporting ETS. The Context Transfer, Handoff Candidate Discovery, and Dormant Mode Host Alerting (SEAMOBY) working group is another relatively new working group in the area of mobile communications. It too is probably too immature at this time to be determined if specific aspects could (or even should) be added to supprt ETS. However, the Carlberg Expires December 9, 2004 [Page 11] Internet Draft ETS Single Domain Framework June 9, 2004 subject area of context transfer is an important one and it has the potential to constructively support ETS. 4.8 Differentiated Service (Diff-Serv) There are a number of examples where Diff-Serv [22] has been deployed strictly within a domain, with no extension of service to neighboring domains. Various reasons exist for Diff-Serv not being widely deployed in an inter-domain context, including ones rooted in the complexity and problems in supporting the security requirements for Diff-Serv code points. An extensive discussion on Diff-Serv deployment is outside the scope of this document. [23] presents common examples of various codepoints used for well known applications. The document does not recommend these associations as being standard or fixed. Rather, the examples in [23] provide a reference point for known deployments that can act as a guide for other network administrators. An arguement can be made that Diff-Serv, with its existing code point specifications of Assured Forwarding (AF) and Expedited Forwarding (EF), goes beyond what could be needed to support ETS within a domain. By this we mean that the complexity in terms of maintenance and support of AF or EF may exceed or cause undue burden on the management resources of a domain. Given this possibility, users or network administrators may wish to determine if various queuing mechansisms, like class based weighted fair queuing, is sufficient to support ETS flows through a domain. Note, as we stated earlier in section 2, over provisioning is another option to consider. We assume that if the reader is considering something like Diff-Serv, then it has been determined that over provisioning is not a viable option given their individual needs or capabilities. 5. Security Considerations Services used to offer better or best available service for a particular set of users (in the case of this document, ETS users) are prime targets for security attacks, or simple misuse. Hence, administrators that choose to incorporate additional protocols/services to support ETS are strongly encouraged to consider new policies to address the added potential of security attacks. These policies, and any additional security measures, should be considered independent of any firewalls that may exist at the edges of the administrative domain. 6. Acknowledgements Thanks to Ran Atkinson for comments and suggestions on the initial Carlberg Expires December 9, 2004 [Page 12] Internet Draft ETS Single Domain Framework June 9, 2004 version of this draft. 7. References 7.1 Normative Reference 1 Bradner, S., "The Internet Standards Process -- Revision 3", BCP 9, RFC 2026, October 1996. 7.2 Informative References 2 Carlberg, K., Atkinson, R., "Requirements for Supporting ETS in Stub Domains", Internet Draft, Work In Progress, June 2003 3 Carlberg, K,. et. al, "Framework for Supporting ETS in IP Telephony", Internet Draft, Work In Progress, June, 2003 4 Perkins, C., "IP Mobility Support for IPv4", RFC 3344, August 2002 5 Droms, R., "Dynamic Host Configuration Protocol", RFC 2131, March 1997 6 Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol Label Switching Architecture", RFC 3031, January 2001. 7 Le Faucheur, F., et al, "MPLS Support of Differentiated Services", RFC 3270, May 2002 8 Braden, R., et al, "Resource Reservation Protocol (RSVP) Version 1 Functional Specification", RFC 2205, September 1997 9 Awduche, D., "RSVP-TE: Extensions to RSVP for LSP Tunnels", RFC 3209, December 2001 10 Herzog, S., "RSVP Extensions for Policy Control", RFC 2750, January 2000 11 Durham, D., et al, "The COPS (Common Open Policy Service) Protocol", RFC 2748, January 2000. 12 Herzog, S., et al, "COPS Usage for RSVP", RFC 2749, January 2000 13 Chan, K., et al, "COPS Usage for Policy Provisioning (COPS-PR)", RFC 3084, March 2001 Carlberg Expires December 9, 2004 [Page 13] Internet Draft ETS Single Domain Framework June 9, 2004 14 Schoenwaelder, J., "Overview of the 2002 IAB Network Management Workshop", RFC 3535, May 2003 15 "Information technology - Telecommunications and information exchange between systems - Local and metropolitan area networks - Common specifications - Part 3: Media Access Control (MAC) Bridges: Revision. This is a revision of ISO/IEC 10038: 1993, 802.1j-1992 and 802.6k-1992. It incorporates P802.11c, P802.1p and P802.12e." ISO/IEC 15802-3:1998" 16 "Data-Over-Cable Service Interface Specifications: Cable Modem to Customer Premise Equipment Interface Specification SP-CMCI- I07-020301", DOCSIS, March 2002, http://www.cablemodem.com. 17 Meyer, D., Fenner, B., "Multicast Source Discovery Protocol (MSDP)", RFC 3618, October 2003 18 Estrin, D., et al, "Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol Specification", RFC 2362, June 1998 19 Guttman, C., et al, "Service Location Protocol, Version 2", RFC 2608, June 1999. 20 Partridge, C., et al, "Host Anycasting Service", RFC 1546, November 1993 21 Hinden, R., Deering, S., "Internet Protocol Version 6 (IPv6) Addressing Architecture", RFC 3513, April 2003 22 Nichols, K., et al, "Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers", RFC 2474, December 1998. 23 Baker, F., et. al., "Configuration Guidelines for DiffServ Service Classes", Internet Draft, Work In Progress, Feb 2004 8. Author's Addresses Ken Carlberg G11 123a Versailles Circle Baltimore, MD USA carlberg@g11.org.uk Full Copyright Statement Carlberg Expires December 9, 2004 [Page 14] Internet Draft ETS Single Domain Framework June 9, 2004 "Copyright (C) The Internet Society (2004). 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 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 as 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 OR MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Carlberg Expires December 9, 2004 [Page 15]