Conex Group D. McDysan Internet Draft Verizon Intended Status: Informational Expires: September 7, 2011 March 7, 2011 Usage/Volume Tier Feedback Use Case for Congestion Exposure draft-mcdysan-conex-volumetier-usecase-00.txt Status of this Memo This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79. 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 This Internet-Draft will expire on April 17, 2011. Copyright Notice Copyright (c) 2011 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. McDysan Expires September 7, 2011 [Page 1] Abstract As requested in the Beijing meeting, this is an individual draft that expands on the usage tier/volume use case from [McDysan]. The feedback recorded in the Beijing conex meeting minutes was that a number of people were potentially interested in this use case, but that it was out of scope of the current charter, and/or could potentially be built out of the conex abstract mechanism. Table of Contents 1. Introduction...................................................2 2. Conventions used in this document..............................3 2.1. Acronyms..................................................3 2.2. Terminology...............................................3 3. Motivation and Background......................................3 4. Usage Tier/ Volume Feedback....................................3 4.1. Problem Statement.........................................4 4.2. Objectives for Addressing this Issue......................4 4.3. Potential Support Using Abstract Mechanism................5 4.4. Additional Support Using other Measures and Mechanisms....5 5. Security Considerations........................................7 6. IANA Considerations............................................7 7. References.....................................................8 7.1. Normative References......................................8 7.2. Informative References....................................8 8. Acknowledgments................................................9 1. Introduction As requested in the Beijing meeting, this is an individual draft that expands on the usage tier/volume use case from [McDysan]. The feedback recorded in the Beijing conex meeting minutes was that a number of people were interested in this idea, but that it was out of scope of the current charter, and/or could potentially be built out of the conex abstract mechanism. Section 3 provides some motivational background and a statement of relevant problems involved with congestion pricing. Section 4 provides text for the usage/volume tier feedback use case. It contains a section that covers a problem statement, objectives for resolving this issue, potential approaches for implementing this use case employing currently defined conex mechanisms, and a description of additional measures and mechanisms that could solve the stated problem and issues. McDysan Expires September 7, 2011 [Page 2] 2. Conventions used in this document 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]. 2.1. Acronyms conex = congestion exposure 2.2. Terminology The following is a quote from the CONEX working group charter: " ... develop a mechanism by which senders inform the network about the congestion encountered by previous packets on the same flow ... at the IP layer, such that the total level of congestion is visible to all IP devices along the path" 3. Motivation and Background This section provides references to relevant presentations given by experts on congestion pricing from the IETF 78 Technical Plenary in Maastricht. A significant number of ISPs implement some form of usage/volume caps in at least some parts of the world [NewAmerica]. The congestion exposure problems addressed in this document are: o Volume-based pricing makes it difficult for users to manage costs incurred, [Varian] o Customers will pay a premium for unmetered use [Varian] There are three time scales over which congestion pricing can operate [Johari]: short (milliseconds to seconds), medium (minutes to hours to days) and long (months to years). Currently, the short term congestion signal is lost packets or a specific indication of congestion of a particular resource (e.g., a ECN indication for queue congestion), as stated in the conex charter, [UseCases] and [Mechanism]. Usage/Volume tier pricing is typically of a long timescale (months), but users may be able to make effective of shorter time scale feedback (minutes to hours). 4. Usage Tier/ Volume Feedback Usage/volume caps may be arranged into multiple tiers with different pricing based upon monthly volume. This results in some problems as described in the next section. Next, high-level objectives to address these issues are then proposed. Then potential ways that already defined means [Mechanisms] may be employed are described. Finally, to address some of these issues, other measures and mechanisms that could possibly better meet the objectives are described. McDysan Expires September 7, 2011 [Page 3] 4.1. Problem Statement Long-term (e.g., monthly) usage/volume based pricing is a widely used incentive, but it creates the following issues; o It is complex for users to keep track of usage and manage their activity to control the price they pay for access [Varian] o It does not address the situation where heavy users [Usecases] send at a high rate, but only for a fraction of the usage measurement interval (e.g., only for a few hours or days during a month). o If usage/volume counting is performed differently dependent upon the degree of congestion experienced, then feedback is more important since in general users will not know when congestion is occurring. o If a user marks packets as requesting lower effort [LowerEffort], then an incentive could be to not count (or count in a different way) packets marked as lower effort against a usage/volume tier. Some way of ensuring that packets marked as lower effort do not significantly impact packets marked as best (or better) effort. 4.2. Objectives for Addressing this Issue Provide a way to inform a receiver of the usage/volume incurred to a moment in time. Ideally, this would also include the usage/volume time period (e.g., a month). Provide a way to inform a receiver of a trend that if usage continues at the same rate then a specific usage/volume tier will be crossed. Indicate to a receiver whether usage/volume counting is occurring in a different way when congestion measure of a particular form (e.g., loss, ECN marking) is occurring. Standardize a way to mark packets in a way (e.g., [Lower Effort]) in conjunction with some form of conex signaling that indicates usage/volume counting will not occur (or are counted separately) under the condition that these packets do not create congestion. A means to ensure that these marked packets do not create congestion and do not impact best (and better) effort marked packets is also required. Enable a means for recharging to occur, where usage/volume counting does not occur for the receiving user since some other party has agreed to incur the cost of usage/volume for that flow. McDysan Expires September 7, 2011 [Page 4] 4.3. Potential Support Using Abstract Mechanism The conex abstract mechanism [Mechanism] defines implicit signaling of loss and explicit signaling of ECN marking. It also defines re- echoed signal for loss and ECN marking based upon feedback carried by TCP from a receiver back to the sender (in an RFC to be developed by the conex wg as defined in the charter). If this feedback mechanism is designed to be extensible, then a variety of forms of feedback could be developed for use in experiments. Counting usage/volume differently for congested packets (or congested intervals) based upon [Mechanism] re-echoed congestion experienced signals seems straightforward. This could be a local matter for the IP node which implements usage/volume tier counting. Also, counting packets marked as Lower Effort differently is a local matter. How to ensure that these packets do not interfere with best effort could be implemented by Diffserv methods locally and at other potential bottlenecks. Since packets subject to counting in a usage/volume cap may not occur during congestion intervals, reinsertion of such counting information using the re-echoed signals that indicate congestion does not seem possible since the same bits cannot represent usage counting and congestion experienced. What is missing from the current conex mechanism is a feed forward path operating over a longer timescale that contains sufficient information to meet the objectives. 4.4. Additional Support Using other Measures and Mechanisms Usage/volume counting has some aspects similar to that of a congestible queue, but on a much longer timescale, as follows: o Instead of a queue which is typically sized for O(10 ms) at the sending rate, usage/volume counting occurs on a timescale O(month. o A usage/volume tier is a threshold on a long term usage counter, similar to the way ECN marking can be a threshold on in a queue. o Queue loss is similar to a usage/volume counter crossing from one tier into the next. o A usage/volume tier trend warning is similar to a rate estimate for ECN marking based upon queue fill rate, as is described in PCN. Therefore, in an abstract way a usage/volume counter can be viewed as a congestible resource, but in some ways not the same as a congestible queue. If this information is to be fed forward in a way observable at the IP layer and fed back at the transport layer (e.g., McDysan Expires September 7, 2011 [Page 5] TCP), then additional packet and transport fields and/or mechanisms may be better suited to this purpose. Furthermore, instead of feeding forward information in each IPv6 packet as in [Mechanism], usage/volume congestible resource information can occur much less frequently (e.g., many minutes to hours). The following is an outline of such measures and mechanisms. The basic idea is based on the fact that the sender and receiver need to be cooperating using the same experimental extensions to TCP, and that if TCP can carry some of the additional information, then the scarcity of IPv6 header bits is avoided. Furthermore, as described previously, "fast path" processing is not required for this use case and the hop-by-options field of the IPv6 header could be used [RFC2460], [IPv6Format] with "slow path" processing used instead. A mechanism to allow the experimental sender to send a "probe" in the IPv6 packet (e.g., using an experimental IPv6 protocol type) could be used by a intermediate IP node(s) to forward the "probe" packet to a special processor (which may be separate from the routers' processor). This special processor could use a polled version of usage/volume count information per user and could also be configured with subscription information (e.g., usage/volume cap tier, cap duration), and threshold settings. The handling of this "probe" IPv6 packet and associated TCP segment needs to be done within the TCP flow. It could use an Out Of Band mechanism similar to the urgent data capability in TCP. (For example, an experimental usage of the Urgent bit could possibly be employed.) The special processor could insert additional measures and implement some of the proposed mechanisms and then modify/augment the "probe" TCP (urgent-like) segment with the requested information and forwards this modified "probe" packet toward the receiver via the intermediate IP node. Packets from the receiver back to the sender could be sent directly, or could be directed through the special processor at intermediate node(s), depending upon the specifics of the use case involved. A consequence of the above extension of measures and mechanisms is that the sender and receiver now have much more information which could be used to solve the stated problems and meet the objectives. The information carried in a "probe" TCP segment could include: o The service being requested, for example: o Request information on the users' usage/volume tier o Request statistics on usage o Request threshold trend report o Request not counting this flow since it is lower effort McDysan Expires September 7, 2011 [Page 6] o Request recharging o Information that could be provided by the "special processor" includes: o Duration and cap for the usage volume measurement tier (e.g., a month) o The absolute count of packets and octets received/sent, and/or fraction of the usage tier already used o Count of packets and octets received/sent which experienced congestion o Count of packet and octets received/sent that were marked as Lower Effort o Estimate of whether the user will exceed the usage tier if the historical usage rate to the reporting instant continues o A pointer (e.g., URL) and identification of the authentication method that would enable other queries, and/or implement alternative charging methods (e.g., recharging) o Other measures related to the "congestion" of a usage/volume tier use case (or possibly other use cases as well). One example of a different type of measure is described in [Stanojevic]. In this paper, the Shapley value [Shapley] is used instead of a 95-th percentile measure of hourly usage measurements across a month. The Shapley value has the following desirable intuitive properties [Shapley]: individual fairness, efficiency, symmetry and additivity. Although the 95-th percentile measure is not directly related to the usage/volume tier use case, the authors state that this is a case they plan to address in future research. A mapping, such as an approximation to the Shapley value described in this paper, could be a way to compress the usage/volume tier feed forward/ feedback information into a smaller number of bits that represents the incentives described in the objectives section. 5. Security Considerations In the proposed mechanisms there are indications that could be spoofed and/or used to game counting and congestion feedback mechanisms, and therefore an authentication mechanism may be needed when this information is handled at the TCP/IPv6 layer in the sender to destination direction or at the TCP layer in the destination to sender direction. 6. IANA Considerations None McDysan Expires September 7, 2011 [Page 7] 7. Acknowledgements The idea of not counting lower effort traffic against a usage/volume cap was suggested my Mikael Abrahamsson on the conex mailing list. 8. References 8.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. 8.2. Informative References [UseCases] B. Briscoe, R. Woundy, T. Moncaster, Ed., J. Leslie, Ed., "ConEx Concepts and Use Cases," draft-moncaster-conex-concepts-uses- 01, Work in Progress [Mechanism] M. Mathis, B. Briscoe, "Congestion Exposure (ConEx) Concepts and Abstract Mechanism," draft-mathis-conex-abstract-mech- 00, Work in Progress [Varian] Hal Varian, Google, "Congestion pricing principles," IETF 78 Technical Plenary, 29 July 2010 [Johari] Ramesh Johari, Stanford University, "The information in congestion prices: milliseconds to years," IETF 78 Technical Plenary, 29 July 2010 [NewAmerica] Li, Losey, "Bandwidth Caps for Residential High- Speed Internet in the U.S. and Japan," August 2009, http://www.newamerica.net/files/Bandwidth%20Caps%20for%20High- Speed%20Internet%20in%20the%20U.S.%20and%20Japan.pdf [LowerEffort] R. Bless, K. Nichols, K. Wehrle, "A Lower Effort Per- Domain Behavior (PDB) for Differentiated Services," RFC3662, December 2003 [RFC2460] S. Deering, R. Hinden, "Internet Protocol, Version 6 (IPv6) Specification," RFC 2460 December 1998 [IPv6Format] S. Krishnan, M. Kuehlewind, "IPv6 Format," Presentation at conex wg at IETF 79, Beijing [Stanojevic] Stanojevic, Laoutaris, Rodriguez, Telefonica Research, "On Economic Heavy Hitters: Shapley value analysis of 95th-percentile pricing," IMC'10, November 1-3, 2010, Melbourne, Australia, http://conferences.sigcomm.org/imc/2010/papers/p75.pdf [Shapley] Wikipedia, "Shapley value," http://en.wikipedia.org/wiki/Shapley_value McDysan Expires September 7, 2011 [Page 8] 9. Acknowledgments This document was prepared using 2-Word-v2.0.template.dot. Copyright (c) 2011 IETF Trust and the persons identified as authors of the code. All rights reserved. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. This code was derived from IETF RFC [insert RFC number]. Please reproduce this note if possible. Authors' Addresses Dave McDysan Verizon 22001 Loudoun County PKWY Ashburn, VA 20147 Email: dave.mcdysan@verizon.com McDysan Expires September 7, 2011 [Page 9]