Conex Group D. McDysan Internet Draft Verizon Intended Status: Informational Expires: April 17, 2011 October 17, 2010 Proposed Additional Use Cases for Congestion Exposure draft-mcdysan-conex-other-usecases-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) 2010 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 April 17, 2011 [Page 1] Abstract This draft proposes some use cases for inclusion in the conex Working group charter's deliverable for an informational RFC covering use case description. These use cases are in addition to and/or complement those described in [UseCases], and focus on forms of congestion exposure that involve resources other than queues and timeframes other than real-time. Table of Contents 1. Introduction...................................................2 2. Conventions used in this document..............................2 2.1. Acronyms..................................................3 2.2. Terminology...............................................3 3. Motivation and Background......................................3 4. Proposed Use Cases.............................................4 4.1. Inequity of Heavy versus Light Users......................4 4.2. Usage Tier/ Volume Feedback...............................4 4.3. Feedback on Time of Day, Day of Week Charging.............5 4.4. Recharging for Implementing Congestion Pricing............6 5. Security Considerations........................................6 6. IANA Considerations............................................6 7. References.....................................................6 7.1. Normative References......................................6 7.2. Informative References....................................7 8. Acknowledgments................................................7 1. Introduction This draft proposes some use cases for inclusion in the conex Working group charter's deliverable for an informational RFC covering use case description. These use cases are in addition to and/or complement those described in [UseCases], and focus on forms of congestion exposure that involve resources other than queues and timeframes other than real-time. Section 3 provides some motivational background and a statement of problems involved with congestion pricing, with references to the presentations by experts in this area at the IETF 78 Technical Plenary in Maastricht. Section 4 provides text for each of the above use cases in a mechanism independent manner. 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]. McDysan Expires April 17, 2011 [Page 2] 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 Successful adoption of experimental conex protocol(s) must address use cases that provide significant value to users, content providers and service providers. Central themes to this value proposition are incentives (i.e., congestion pricing) and the cost of providing marginal capacity [Varian, Johari]. 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. Setting congestion price as the marginal value of capacity is useful for medium timescales via traffic engineering and longer time scales via provisioning [Johari]. Some congestion exposure problems are challenging to address and are not completely addressed in [UseCases]: o 20% of the users generate 80% of the traffic and create unfairness with certain resource sharing [Varian] o Volume-based pricing makes it difficult for users to manage costs incurred, [Varian], [Briscoe] o Customers will pay a premium for unmetered use [Varian] o A form of congestion pricing is "recharging" (e.g., "free shipping") [Varian], where someone other than the end user pays for incurred congestion. o In general a content or service provider has hard capacity constraints at certain bottlenecks in their infrastructure (e.g., server capacity, router interface/queue service rate). Some form of adaption, such as time-shifting, route-shifting, or moderating the demand is required to adapt to these constraints [Kelly]. McDysan Expires April 17, 2011 [Page 3] If conex exposes congestion without damage (e.g., loss) then many forms of adaption are feasible, as long as incentives are aligned with the signaled congestion [Kelly]. The use cases in the next section focus on forms of adaption that enable certain sets of incentives that are not completely covered in [UseCases]. 4. Proposed Use Cases 4.1. Inequity of Heavy versus Light Users In many networks, 20% are heavy users generating 80% of the traffic [Varian]. This means that a heavy user generates 16 times the traffic as a light user as a medium term (e.g, monthly) average. But, in a bandwidth-tiered flat priced network, heavy and light users often pay nearly the same price since pricing is based upon the short term (milliseconds) bandwidth measure of a shaper and/or a policer. During non-peak periods, the resources of a service or content provider are underutilized and the marginal cost of capacity is small. The access network is provisioned and traffic engineered for peak capacity of all users, and when congested, heavy users create 16 times the congestion of small users. However, during peak use periods, a heavy user may send at near the bandwidth tier while light users may send intermittently. There is a need for a means for service providers to equitably assign costs to heavy versus light users. For example, the light users may pay less if they were charged by volume, as described in another use case. A congestion measure of burstiness (e.g., ratio of peak rate to average rate over a longer interval than the tiered bandwidth shaper or policer) could be helpful in this use case. In general, a bursty packet flow is light (e.g., web surfing) and a non-bursty packet flow is heavy (e.g., viewing a lengthy HD video). The destination could perform this measure and feed it back to the sender. It may only be necessary to feedback this measure for heavy users, since the absence of such feedback could be inferred as an indication of a light user. The sender could insert some processed version of this feedback measure this at the IP layer so that all IP devices could be aware of whether this is a heavy or light user. 4.2. Usage Tier/ Volume Feedback Long-term (e.g., monthly) usage volume based pricing can more equitably assign costs to the prices paid, but; o is complex for users to keep track of usage and manage their activity to control the price they pay for access [Varian], [Briscoe], McDysan Expires April 17, 2011 [Page 4] o does not address situation where heavy users 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 counting is performed differently dependent upon the amount of congestion incurred (i.e., some form of congestion pricing as an incentive), then feedback is more important since in general users will not know when congestion is occurring, and even if they were informed this makes their usage tracking problem even more complex. If usage volume information could be fed forward from an IP device using a conex mechanism on the path from senders to a destination, then this information could be fed back from the destination to the sender using TCP as stated in the Conex charter. Such information could include: the duration of the usage volume measurement tier (e.g., a month), the fraction of the usage tier already used, estimate of whether the user will exceed the usage tier if the historical rate to date continues, and any other information (or a pointer to such information) that would address the challenges of usage tier based pricing. There are instances where a service/ content provider may choose to not count certain packets against the volume tier, such as when there is no congestion occurring, recharging is being done for this packet flow, and/or there is no usage counting being done for this time of day/week. A secure means to feed such information forward at the IP layer would allow for a number of different forms of counting, and hence adaption to congestion to occur. 4.3. Feedback on Time of Day, Day of Week Charging Congestion occurs when the offered load approaches that of the provisioned capacity, which often does not occur until shortly before there would be a need to provision additional capacity. Depending upon how restoration capacity is allocated by a service/content provider, congestion may only occur during peak periods when a failure is present. Without Conex, utilization averaged over several minutes can be as high as 70 to 80% in typical network bottlenecks without loss that would reduce TCP effective throughput (i.e., goodput). A short term congestion control (sub-second to seconds) method that could increase utilization to above 90% and still avoid loss would only increase effective capacity by 10-20%. If traffic increases at 50-75% per year, then a 10-20% increase in effective capacity means that the provisioning interval is only shifted by a few months. This handling of short term congestion use case alone may not be sufficient motivation for a service provider to deploy congestion handling measures. McDysan Expires April 17, 2011 [Page 5] However, in many points of a network, the majority of usage occurs during peak periods (e.g, a few busy hours) while much spare capacity exists off peak. The product of the spare capacity (bits/second) and the non-peak interval (seconds) that could carry traffic ranges from 2 to 10 times of the traffic carried during the peak period. Congestion exposure and congestion pricing that enables users and content providers to time shift traffic to off-peak periods that would have otherwise been sent during peak periods can reduce provisioned capacity cost by as much as several hundred percent. Historical time of day usage patterns could be employed to time shift traffic, but often maintenance actions are performed during the off peak periods, making the prediction of congestion using these methods less reliable. An automatic method for detection of congestion during off-peak periods is highly desirable. 4.4. Recharging for Implementing Congestion Pricing There should be a means to recharge (i.e., someone other than the receiving user pays) for usage that causes congestion during peak demand period versus that which does not [Varian]. If TCP were augmented with information related to the form of congestion, including not only short term as covered in [UseCases], but also including usage tier, Time of Day, or burstiness then sufficient information to implement sender pays (e.g., Content provider)versus receiver pays (e.g., end user) could be implemented. When the sender includes this information in the IP layer, then usage tier counting and TOD counting could be accounted for differently in IP devices in the path between sender and destination. Such an indication of recharging would need be to be authenticated in some way. 5. Security Considerations Some use cases involve indications that could be spoofed or used to game counting and congestion feedback mechanisms, and therefore an authentication mechanism is needed when this information is handled at the IPv6 layer in the sender to destination direction or at the TCP layer in the destination to sender direction. 6. IANA Considerations None 7. References 7.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. McDysan Expires April 17, 2011 [Page 6] 7.2. 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 [Varian] Hal Varian, Google, "Congestion pricing principles," IETF 78 Technical Plenary, 29 July 2010 [Kelly] Frank Kelly, University of Cambridge, "Economic perspectives on congestion," 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 [Briscoe] Bob Briscoe, BT, "Congestion Exposure," IETF 78 Technical Plenary, 29 July 2010. 8. Acknowledgments This document was prepared using 2-Word-v2.0.template.dot. Copyright (c) 2010 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 April 17, 2011 [Page 7]