TSVWG R. Geib, Ed. Internet-Draft Deutsche Telekom Intended status: Informational February 25, 2013 Expires: August 29, 2013 DiffServ interconnection classes and practice draft-geib-tsvwg-diffserv-intercon-02 Abstract This document proposes a limited set of interconnection QoS PHBs and PHB groups. It further introduces some DiffServ deployment aspects. The proposals made here should be integrated into a revised version of RFC5127. Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. 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." This Internet-Draft will expire on August 29, 2013. Copyright Notice Copyright (c) 2013 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Geib Expires August 29, 2013 [Page 1] Internet-Draft Abbreviated Title February 2013 This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 4 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 3. An Interconnection class and codepoint scheme . . . . . . . . 5 4. Consolidation of QoS standards by the interconnection codepoint scheme . . . . . . . . . . . . . . . . . . . . . . . 6 5. MPLS, Ethernet and IP Precedence for aggregated classes . . . 8 6. QoS class name selection . . . . . . . . . . . . . . . . . . . 9 7. Allow for DiffServ extendability on MPLS and Ethernet level . 10 8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 10 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 10. Security Considerations . . . . . . . . . . . . . . . . . . . 10 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 10 11.1. Normative References . . . . . . . . . . . . . . . . . . . 10 11.2. Informative References . . . . . . . . . . . . . . . . . . 11 Appendix A. Change log . . . . . . . . . . . . . . . . . . . . . 12 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 12 Geib Expires August 29, 2013 [Page 2] Internet-Draft Abbreviated Title February 2013 1. Introduction This draft proposes a DiffServ interconnection class and codepoint scheme. At least one party of an interconnection often is a network provider. Aggregated DiffServ classes are often deployed within provider networks. To respect this, this draft also contains concepts and current practice relevant for a revised version of RFC5127 [RFC5127]. Its main purpose is to be considered as an input for the latter task. DiffServ sees deployment in many networks for the time being. As described in the introduction of the draft diffserv problem statement [I-D.polk-tsvwg-diffserv-stds-problem-statement], remarking of packets at domain boundaries is a DiffServ feature. This draft proposes a set of standard QoS classes and codepoints at interconnection points to which and from which locally used classes and codepoints should be mapped. Such a scheme simplifies interconnection negotiations and ensures that end to end class properties remain roughly the same, even if codepoints change. The proposed Interconnection class and codepoint scheme tries to reflect and consolidate related DiffServ and QoS standardisation efforts outside of the IETF, namely MEF, GSMA and ITU. IP Precedence has been depricated when DiffServ was standardised. It is common practice today however to copy the DSCPs "IP Precedence Bits" into MPLS TC or Ethernet P-Bits, whenever possible. This is reflected by the DiffServ codepoint definitions of AF and EF. This practice and it's limits deserve to be documented and disussed briefly. The draft further adds proposes a philosophy how to add or pick aggregated DiffServ classes. The set of available router and traffic management tools to configure and operate DiffServ classes is limited. This should be reflected by class definitions. These may in the end be more related to transport properties than to application requirements. Please interpret transport properties as "congestion aware" and "not congestion aware" rather then TCP or UDP. Finally, this draft proposes to leave some MPLS TC codepoint space to allow for future DiffServ extensions like ECN/PCN and domain internal classes (network management traffic is a good example for the latter). An example for an internal PHB may be CS6, which some operators to protect their ntework internal routing and / or management traffic. This PHB may not be available to transport customer signaling and management traffic. It IETF is interested in this work, a later version may expand on internal PHBs and codepoints. Geib Expires August 29, 2013 [Page 3] Internet-Draft Abbreviated Title February 2013 In addition to the standardisation activities which triggered this work, other authors published RFCs or drafts which may benefit from an interconnection class- and codepoint scheme. RFC 5160 suggests Meta-QoS-Classes to enable deployment of standardised end to end QoS classes [RFC5160]. The authors agree that the proposed interconnection class- and codepoint scheme as well as the idea of standardised end to end. Hence RFC 5160 and this work complement each other. Work on BGP Class of Service Interconnection signaled by BGP [I-D.knoll-idr-cos-interconnect] is beyond the the scope of this draft. Should the basic transport and class properties of end to end QoS utilising DiffServ based interconnection as proposed by this draft be standardised, work on signaled access to QoS classes may be of interest. 1.1. Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119]. 2. Terminology This draft tries to re-use exitsing terminology. It further tries to indicate, where duplicate terminology may exist. Class A class is a set of one or more PHBs. If a class consists of a set of PHBs and these obey to an ordering constraint. In that sense, a class is a single AF class (e.g. AF4 consisting of AF41, AF41 and AF43) [RFC2597]. A class is a PHB group [RFC2575] and a PHB scheduling class [RFC3260]. On IP level all DSCPs sharing the same IP precedence value belong to a single class. A class may consist of one or more PHBs. A single class uses forwarding resources, which are independent of the forwarding rseources of any other class. Different classes must not be aggregated. PHB A single Per Hop Behaviour [RFC2575] is identified by a single DSCP on IP layer. Many DiffServ related RFCs introduce new terminolgy duplicating the existing one. The above references are incomplete and refer to the early DiffServ RFCs only. Stopping terminology duplication may simplify discussion. The following current practice issues relate to the concept of the DiffServ interconnection class propsal rather than to terminology. They serve as additional motivation of this activity: Geib Expires August 29, 2013 [Page 4] Internet-Draft Abbreviated Title February 2013 o Abstract class names like "EF" are preferrential over those being close to an application, like "Voice". Unfortunately, non QoS experts can't handle abstract class names. Hence and usually sooner than later, classes are named for applications or groups of them. One consequence however is, that people tend to combine application group class names and SLA parameters. Based on an application specific name and some worst case performance numbers on a paper, they often decide that their application needs a separate new QoS class. o Worse than that, but very present in practice, is the class abstraction level which is preferred by those dealing with QoS (as experts or non experts): the DSCPs or the IP precedence values. These are the commodity abstractions applied for QoS classes. Most of these persons have fixed class to codepoint mappings in their minds, which they can't easily adapt on a per customer or interconnection partner basis. While these issues aren't to be solved by IETF (QoS experts could and should of course teach staff to use proper Diffserv terminology and concepts), a simple and comprehensible QoS interconnection class scheme also is helpful in this area. 3. An Interconnection class and codepoint scheme DiffServ deployments mostly follow loose class specification schemes (often one or two AF classes, EF and Best Effort). Especially DSCP assignment for the AF classes varies between deployments. Basic AF class definitions are often similar however. This is in line with the DiffServ architecture. This document doesn't propose to change that. Interconnecting parties face the problem of matching classes to be interconnected and then to agree on codepoint mapping. As stated by draft diffserv prolebm statement [I-D.polk-tsvwg-diffserv-stds-problem-statement], remarking is a standard behaviour at interconnection interfaces. This draft propses a set of 4 QoS classes with a set of well defined DSCPs and IP- Precedence values as interconnection class and codepoint scheme. A sending party remarks DSCPs from internal schemes to the Interconnection codepoints. The receiving party remarks IP- Precedence and or DSCPs to their internal scheme. Thus the interconnection codepoint scheme fully complies with the DiffServ architecture. Such an interconnection class and codepoint scheme was introduced by ITU-T [Y.1566] (there also includuing Ethernet). It is specified to a higher level of detail in this document. Geib Expires August 29, 2013 [Page 5] Internet-Draft Abbreviated Title February 2013 At first glance, this looks like an additional effort. But there are obvious benefits: each party sending or receiving traffic has to specify the mapping from or to the interconnection class and codepoint scheme only once. Without it, this is to be negotiated per interconnection party individually. Further, end-to-end QoS in terms of traffic being classified for the same class in all passed domains is likely to result if an interconnection codepoint scheme is used. It is not necessarily resulting from individual per network mapping negotiations. The standards and deployments known to the author of this draft are limited to 4 DiffServ classes at interconnection points (or less).Draft RFC 4597 update [I-D.polk-tsvwg-rfc4594-update]doesn't seem to generally contradict to this, as it proposes to standardise "many services classes, not all will be used in each network at any period of time." Some more good reasons fovour working with 4 DiffServ interconnection classes for now: o There should be a coding reserve for interconnection classes, leaving space for future standards, for bilateral agreements and for carrier internal classes. o MPLS and Ethernet support only 8 PHBs, classes or ECN indications. Assignment of codepoints for whatever purpose must be well thought through. Limiting interconnection QoS to four classes is MPLS and Ethernet friendly in that sense. o Migrations from one codepoint scheme to another may require spare QoS codepoints. 4. Consolidation of QoS standards by the interconnection codepoint scheme The interconnection class and codepoint scheme proposed by Y.1566 also tries to consolidate related DiffServ and QoS standardisation efforts outside of the IETF [Y.1566]. The interconnection class and codepoint scheme may be a suitable approach to consolidate these standards. MEF 23.1 specifies 3 aggregated classes, consuming up to 5 codepoints on Ethernet layer (EF, AF3 and AF1 and Best Effort) and 6 PHBs [MEF23.1]. MEF aggregates AF1 and Default PHB in a single class. This is not recommended for interconnection, as it is not in line with RFC 2597 (which requires separate forwarding resources for each AF class and doesn't forsee aggregation of Default PHB and an AF class). GSMA IR.34 proposes four classes, EF, AF4, another AF class and Best Effort with 7 PHBs in sum [IR.34]. IR.34 specifies an "Interactive" Geib Expires August 29, 2013 [Page 6] Internet-Draft Abbreviated Title February 2013 class consisting of 3 PHBs with fifferent drop priorities. IR.34 specfies the PHBS AF31, AF21 and AF11 for this Interactive class. This definitely breaks RFC 2597. The interconnection class and codepoint scheme supports the Interactive class but assigns AF3 with PHBs AF31, AF32 and AF33. If IETF picks up this draft, it may be a good idea to inform MEF and GSMA about conflicts of their standards with DiffServ and suggest joint activities to improve the situation. Information on interworkings with MEF 23 and GSMA IR.34 with the interconnection QoS scheme could be given by a later version of this draft. The classes to be supported at interconnection interfaces are specified by Y.1566 as: Class Priority: EF, expecting the figures of merit describing the PHB to be in the range of low single digit milliseconds. See [RFC3246]. Bulk inelastic: Optimised for low loss, low delay, low jitter at high bandwidth. Traffic load in this class must be controlled, e.g. by application servers. One example could be flow admission control. There may be infrequent retransmissions requested by the application layer to mitigate low levels of packet losses. Discard of packets through active queue management should be avoided in this class. Congestion in this class may result in bursty packet loss. If used to carry multimedia traffic, it is recommended to carry audio and video traffic in a single PHB. All of these properties influence the buffer design. Assured: This class may be optimised to transport traffic without bandwidth requirements. It aims onVery low loss at high bandwidths. Retransmissions after losses characterise the class and influence the buffer design. Active queue management with probabilistic dropping may be deployed. Default: Default. This class may be optimised to transport traffic without bandwidth requirements. Retransmissions after losses characterise the class and influence the buffer design. Active queue management with probabilistic dropping may be deployed. Note that other DiffServ related standards trim down class requirements to SLA parameters. To quote e.g RFC 4594-update, "A "service class" represents a similar set of traffic characteristics for delay, loss, and jitter as packets traverse routers in a network." This draft adds traffic conditioning properties Geib Expires August 29, 2013 [Page 7] Internet-Draft Abbreviated Title February 2013 corresponding to expected transport layer characteristics as a key factor to a class definition: the desired class performance like delay, jitter and worst case loss are met only if conditioning and and transport properties meet the ones described by the class definition. This is not to say, the other standards ignore conditioner properties. They are e.g. a core part of RFC 4594- update. They do not directly refer to tranport protocol properties, as most existing QoS standards prefer the approach of assigning QoS classes to applications or application sets. This may result in undesirable class mappings, if an e.g. IP TV application demanding low loss is matched to a class whose low loss guarantees depend on AQM mechanisms. Y.1566 does not recommend all PHBs to be supported at an interconnection interface. This information is added by this draft. At interconnection points, the following PHBs should be accepted between interconnected parties: Class: PHB (one or more) Class Priority: EF Bulk inelastic: AF41 (AF42 and AF43 are reserved for extension) Assured: AF31, AF32 and AF33 Default: Default Class names (and property specification) are picked from Y.1566. PHBs to the level of detail introduced here are not part of Y.1566. 5. MPLS, Ethernet and IP Precedence for aggregated classes IP Precedence has been depricated when DiffServ was standardised. Ethernet and MPLS support 3 bit codepoint fields to differntiate service quality. Mapping of the IP precedence to these 3 Bit fields has been a configuration restriction in the early days of DiffServ. The concept of paying attention to the three most significant bits of a DSCP has however been part of Diffserv from start on (EF's IP Precedence is 5, that of AF4 is 4 and so on). The interconnection class and codepoint scheme respects this in different ways: o it allows to classify four interconnection classes based on IP precedence. o It supports a single PHB group (AF3), which may be mapped to up to three different MPLS TC's or Ethernet P-Bits. Note that this Geib Expires August 29, 2013 [Page 8] Internet-Draft Abbreviated Title February 2013 draft doesn's favour or recommend doing that, but it is possible. The author isn't aware of deployed service offers with 3 different drop levels in a single class. This is of course no requirement to depricate any DSCP to MPLS TC or Ethernet P-Bit mapping functionality. This functionality is very important as well. 6. QoS class name selection This is more of an informational discussion, proposed best practice, and mainly relates to human behviour (including QoS experts) rather than technical issues. Above the human preference for conceivable class names has been mentioned. Network engineers (including the former Diffserv WG authors) recommend to avoid application related QoS class names. Focus should be put on class properties. But these can be irritating again, as just looking at SLA parameters like Delay, Jitter and packet loss don't tell the reader, which conditioning and transport properties guideed the class engineering assumptions resulted in the conditioning of a class. A router produces QoS with a scheduling mechanism, a settable queue depth and optional active queue management (including ECN), and may be a policer. Some kind of resource management may be present (also in Diffserv domains). It's beyond the imagination of the author how one would engineer more than half a dozen classes with distinguishable properties with this set of tools. There's no perfect solution to the problem, as conditioning configurations are not comprehensible to most readers, even if they were communicated (they are operational secrets of course). There are (or should be) engineering assumptions, when designing QoS conditioners. But they closer relate to layer 3 or layer 4 level properties than to specific applications. In general, an application responds to congestion by reducing traffic, or it ignores congestion. Active queue management doesn't help to avoid congestion in the latter case, only resource management does. EF may be a special case. If the EF traffic is not responsive to congestion, and packets are assumed to be short, rather small jitter values can be reached if enginieering ensures that the packet arrival rate never exceeds the transmision rate of that queue (see RFC 3246 [RFC3246]). There's other non congestion-responsive traffic, for which the EF engineering assumptions may not fit. So conditioning) like bulk inelastic is reasonable. Active queue management may be deployed for QoS classes, which are designed to transport traffic responding to congestion by traffic reduction. Geib Expires August 29, 2013 [Page 9] Internet-Draft Abbreviated Title February 2013 The class names of this document follow Y.1566. TCP_optimised and especially UDP_optimised are inappropriate as clas names, as some UDP based application are or may be expected to become TCP friendly. 7. Allow for DiffServ extendability on MPLS and Ethernet level Any aggregated Diffserv deployment faces codepoint depletion issues rather soon, if deployed on MPLS or Ethernet. Coding space should be left for new features, like ECN, PCN or Conex. In addition to carrying customer traffic, internal routing and network management traffic may be protected by using a separate class. Offering interconnection with up to four classes and 4 - 6 MPLS TC's (or Ethernet P-bits) to that respect is probably at least a fair compromise. 8. Acknowledgements David Black gave many helpful comments to this work. Al Morton and Sebastien Jobert provided feedback on many aspects during private discussions. Brian Carpenter, Mohamed Boucadair and Thomas Knoll helped adding awareness of further potentially related work. 9. IANA Considerations This memo includes no request to IANA. 10. Security Considerations This document does not introduce new features, it describes how to use existing ones. The security section of RFC 4597 [RFC4597] applies. 11. References 11.1. Normative References [RFC2575] Wijnen, B., Presuhn, R., and K. McCloghrie, "View-based Access Control Model (VACM) for the Simple Network Management Protocol (SNMP)", RFC 2575, April 1999. [RFC2597] Heinanen, J., Baker, F., Weiss, W., and J. Wroclawski, "Assured Forwarding PHB Group", RFC 2597, June 1999. Geib Expires August 29, 2013 [Page 10] Internet-Draft Abbreviated Title February 2013 [RFC3246] Davie, B., Charny, A., Bennet, J., Benson, K., Le Boudec, J., Courtney, W., Davari, S., Firoiu, V., and D. Stiliadis, "An Expedited Forwarding PHB (Per-Hop Behavior)", RFC 3246, March 2002. [RFC3260] Grossman, D., "New Terminology and Clarifications for Diffserv", RFC 3260, April 2002. [min_ref] authSurName, authInitials., "Minimal Reference", 2006. 11.2. Informative References [I-D.knoll-idr-cos-interconnect] Knoll, T., "BGP Class of Service Interconnection", draft-knoll-idr-cos-interconnect-09 (work in progress), November 2012. [I-D.polk-tsvwg-diffserv-stds-problem-statement] Polk, J., "The Problem Statement for the Standard Configuration of DiffServ Service Classes", draft-polk-tsvwg-diffserv-stds-problem-statement-00 (work in progress), July 2012. [I-D.polk-tsvwg-rfc4594-update] Polk, J., "Standard Configuration of DiffServ Service Classes", draft-polk-tsvwg-rfc4594-update-02 (work in progress), October 2012. [IR.34] GSMA Association, "IR.34 Inter-Service Provider IP Backbone Guidelines Version 7.0", GSMA, GSMA IR.34 http:/ /www.gsma.com/newsroom/wp-content/uploads/2012/03/ ir.34.pdf, 2012. [MEF23.1] MEF, "Implementation Agreement MEF 23.1 Carrier Ethernet Class of Service Phase 2", MEF, MEF23.1 http:// metroethernetforum.org/PDF_Documents/ technical-specifications/MEF_23.1.pdf, 2012. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC4597] Even, R. and N. Ismail, "Conferencing Scenarios", RFC 4597, August 2006. [RFC5127] Chan, K., Babiarz, J., and F. Baker, "Aggregation of Diffserv Service Classes", RFC 5127, February 2008. [RFC5160] Levis, P. and M. Boucadair, "Considerations of Provider- Geib Expires August 29, 2013 [Page 11] Internet-Draft Abbreviated Title February 2013 to-Provider Agreements for Internet-Scale Quality of Service (QoS)", RFC 5160, March 2008. [Y.1566] ITU-T, "Quality of service mapping and interconnection between Ethernet, IP and multiprotocol label switching networks", ITU, http://www.itu.int/rec/T-REC-Y.1566-201207-I/en, 2012. Appendix A. Change log 00 to 01 Added terminology and references. Added details and information to interconnection class and codepoint scheme. Editorial changes. 01 to 02 Added some references regarding rekated work. Clarified class definitions. Further editorial improvements. Author's Address Ruediger Geib (editor) Deutsche Telekom Heinrich Hertz Str. 3-7 Darmstdadt, 64297 Germany Phone: +49 6151 5812747 Email: Ruediger.Geib@telekom.de Geib Expires August 29, 2013 [Page 12]