<?xml version="1.0"encoding="UTF-8"?>encoding="utf-8"?> <!DOCTYPE rfc SYSTEM"rfc2629.dtd" [ <!-- One method to get references from the online citation libraries. There has to be one entity for each item to be referenced. An alternate method (rfc include) is described in the references. --> <!ENTITY RFC2119 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"> <!ENTITY RFC3124 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3124.xml"> <!ENTITY RFC5348 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5348.xml"> <!ENTITY RFC7478 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7478.xml"> <!ENTITY RFC7656 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7656.xml"> <!ENTITY RFC8087 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8087.xml"> <!ENTITY RFC8382 SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8382.xml"> <!ENTITY I-D.narten-iana-considerations-rfc2434bis SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.narten-iana-considerations-rfc2434bis.xml"> <!ENTITY I-D.ietf-rmcat-nada SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.draft-ietf-rmcat-nada-11.xml"> <!ENTITY I-D.ietf-rmcat-gcc SYSTEM "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.draft-ietf-rmcat-gcc-02.xml"> <!ENTITY I-D.ietf-rmcat-eval-test PUBLIC "" "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-eval-test.xml"> <!ENTITY I-D.ietf-rtcweb-overview PUBLIC "" "http://xml2rfc.tools.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rtcweb-overview.xml"> ]> <?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?> <?rfc strict="yes" ?> <?rfc toc="yes"?> <?rfc tocdepth="3"?> <?rfc symrefs="yes"?> <?rfc sortrefs="yes" ?> <?rfc compact="yes" ?> <?rfc subcompact="no" ?> <?xml-stylesheet type="text/xsl" href="rfc2629.xslt"?>"rfc2629-xhtml.ent"> <rfc xmlns:xi="http://www.w3.org/2001/XInclude" number="8699" category="exp" consensus="true" docName="draft-ietf-rmcat-coupled-cc-09"ipr="trust200902">ipr="trust200902" obsoletes="" updates="" submissionType="IETF" xml:lang="en" tocInclude="true" symRefs="true" sortRefs="true" version="3"> <!--***** FRONT MATTER *****xml2rfc v2v3 conversion 2.33.0 --> <front> <title>Coupledcongestion controlCongestion Control for RTPmedia</title>Media</title> <seriesInfo name="RFC" value="8699"/> <seriesInfo name="Internet-Draft" value="draft-ietf-rmcat-coupled-cc-09"/> <author fullname="Safiqul Islam"initials="S.I."initials="S." surname="Islam"> <organization>University of Oslo</organization> <address> <postal> <street>PO Box 1080 Blindern</street> <code>N-0316</code> <city>Oslo</city><region></region><region/> <country>Norway</country> </postal> <phone>+47 22 84 08 37</phone> <email>safiquli@ifi.uio.no</email> </address> </author> <author fullname="Michael Welzl"initials="M.W."initials="M." surname="Welzl"> <organization>University of Oslo</organization> <address> <postal> <street>PO Box 1080 Blindern</street> <code>N-0316</code> <city>Oslo</city><region></region><region/> <country>Norway</country> </postal> <phone>+47 22 85 24 20</phone> <email>michawe@ifi.uio.no</email> </address> </author> <author fullname="Stein Gjessing"initials="S.G."initials="S." surname="Gjessing"> <organization>University of Oslo</organization> <address> <postal> <street>PO Box 1080 Blindern</street> <code>N-0316</code> <city>Oslo</city><region></region><region/> <country>Norway</country> </postal> <phone>+47 22 85 24 44</phone> <email>steing@ifi.uio.no</email> </address> </author> <dateyear="2019" />month="January" year="2020"/> <area>Transport</area> <workgroup>RTP Media Congestion Avoidance Techniques (rmcat)</workgroup> <keyword>tcp</keyword> <abstract> <t>When multiplecongestion controlledcongestion-controlled Real-time Transport Protocol (RTP) sessions traverse the same network bottleneck, combining their controls can improve the total on-the-wire behavior in terms of delay,lossloss, and fairness. This document describes such a method for flows that have the same sender, in a way that is as flexible and simple as possible while minimizing theamountnumber of changes needed to existing RTP applications.ItThis document also specifies how to apply the method for the Network-Assisted Dynamic Adaptation (NADA) congestion controlalgorithm,algorithm and provides suggestions on how to apply it to other congestion control algorithms.</t> </abstract> </front> <middle> <sectiontitle="Introduction" anchor='sec-intro'>anchor="sec-intro" numbered="true" toc="default"> <name>Introduction</name> <t>When there is enough data to send, a congestion controller attempts to increase its sending rate until the path's capacity has been reached. Some controllers detect path capacity by increasing the sending rate further, until packets are ECN-marked <xreftarget="RFC8087"/>target="RFC8087" format="default"/> or dropped, and then decreasing the sending rate until that stops happening. This process inevitably creates undesirable queuing delay when multiple congestion-controlled connections traverse the same network bottleneck, and each connection overshoots the path capacity as it determines its sending rate. </t> <t>The Congestion Manager (CM) <xreftarget="RFC3124"/>target="RFC3124" format="default"/> couples flows by providing a single congestion controller. It is hard to implement because it requires an additional congestion controller and removes all per-connection congestion control functionality, which is quite a significant change to existingRTP basedRTP-based applications. This document presents a method to combine the behavior of congestion control mechanisms that is easier to implement than the Congestion Manager <xreftarget="RFC3124"/>target="RFC3124" format="default"/> and also requireslessfewer significant changes to existingRTP basedRTP-based applications. It attempts to roughly approximate the CM behavior by sharing information between existing congestion controllers. It is able to honor user-specified priorities, which is required byrtcwebWebRTC <xreftarget="I-D.ietf-rtcweb-overview"/>target="I-D.ietf-rtcweb-overview" format="default"/> <xreftarget="RFC7478"/>.</t>target="RFC7478" format="default"/>.</t> <t>The described mechanisms are believed safe to use, but they are experimental and are presented for wider review and operational evaluation.</t> </section> <sectiontitle="Definitions" anchor='sec-def'>anchor="sec-def" numbered="true" toc="default"> <name>Definitions</name> <t>The key words"MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY","<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>", "<bcp14>MAY</bcp14>", and"OPTIONAL""<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as described in BCP 14 <xreftarget="RFC2119">RFC 2119</xref>.</t> <t><list style="hanging" hangIndent="6"> <t hangText="Available Bandwidth:"> <vspace />target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they appear in all capitals, as shown here.</t> <dl newline="true" spacing="normal" indent="6"> <dt>Available Bandwidth:</dt> <dd> The available bandwidth is the nominal link capacity minus the amount of traffic that traversed the link during a certain time interval, divided by that timeinterval.</t> <t hangText="Bottleneck:"> <vspace />interval.</dd> <dt>Bottleneck:</dt> <dd> The first link with the smallest available bandwidth along the path between a sender andreceiver.</t> <t hangText="Flow:"> <vspace />receiver.</dd> <dt>Flow:</dt> <dd> A flow is the entity that congestion control is operating on. It could, for example, be atransport layer connection,transport-layer connection or an RTP stream <xreftarget="RFC7656"/>,target="RFC7656" format="default"/>, regardless of whether or not this RTP stream is multiplexed onto an RTP session with other RTPstreams.</t> <t hangText="Flowstreams.</dd> <dt>Flow Group Identifier(FGI):"> <vspace />(FGI):</dt> <dd> A unique identifier for each subset of flows that is limited by a commonbottleneck.</t> <t hangText="Flowbottleneck.</dd> <dt>Flow State Exchange(FSE):"> <vspace />(FSE):</dt> <dd> The entity that maintains information that is exchanged betweenflows.</t> <t hangText="Flowflows.</dd> <dt>Flow Group(FG):"> <vspace />(FG):</dt> <dd> A group of flows having the sameFGI.</t> <t hangText="SharedFGI.</dd> <dt>Shared Bottleneck Detection(SBD):"> <vspace />(SBD):</dt> <dd> The entity that determines which flows traverse the same bottleneck in thenetwork,network or the process of doingso.</t> </list></t>so.</dd> </dl> </section> <sectiontitle="Limitations" anchor='sec-limits'> <t><list style="hanging" hangIndent="6"> <t hangText="Sender-side only:"> <vspace />anchor="sec-limits" numbered="true" toc="default"> <name>Limitations</name> <dl newline="true" spacing="normal" indent="6"> <dt>Sender-side only:</dt> <dd> Shared bottlenecks can exist when multiple flows originate from the samesender,sender or when flows from different senders reach the same receiver (see <xreftarget="RFC8382"/>, section 3).target="RFC8382" sectionFormat="of" section="3" />). Coupled congestioncontrolcontrol, as describedherehere, only supports the former case, not the latter, as it operates inside a single host on the sender side.</t> <t hangText="Shared</dd> <dt>Shared bottlenecks do not changequickly:"> <vspace />quickly:</dt> <dd> As per the definition above, a bottleneck depends on cross traffic, and since such traffic can heavily fluctuate, bottlenecks can change at a high frequency (e.g., there can be oscillation between two or more links). This means that, when flows are partially routed along different paths, they may quickly change between sharing and not sharing a bottleneck. For simplicity, here it is assumed that a shared bottleneck is valid for a time interval that is significantly longer than the interval at which congestion controllers operate. Note that, for the only SBD mechanism defined in this document (multiplexing on the same five-tuple), the notion of a shared bottleneck stays correct even in the presence of fast trafficfluctuations:fluctuations; since all flows that are assumed to share a bottleneck are routed in the same way, if the bottleneck changes, it will still beshared.</t> </list></t>shared.</dd> </dl> </section> <sectiontitle="Architectural overview" anchor='sec-arch'> <t>Figure 1anchor="sec-arch" numbered="true" toc="default"> <name>Architectural Overview</name> <t><xref target="fig_1"/> shows the elements of the architecture for coupled congestion control: the Flow State Exchange (FSE), Shared Bottleneck Detection(SBD)(SBD), and Flows. The FSE is a storage element that can be implemented in two ways: active and passive. In the active version, it initiates communication with flows and SBD. However, in the passive version, it does not actively initiate communication with flows and SBD; its only active role is internal state maintenance (e.g., an implementation could use soft state to remove a flow's data after long periods of inactivity). Every time a flow's congestion control mechanism would normally update its sending rate, the flow instead updates information in the FSE and performs a query on the FSE, leading to a sending rate that can be different from what the congestion controller originally determined. Using information about/from the currently active flows, SBD updates the FSE with the correct Flow Group Identifiers (FGIs). </t> <t> This document describes both active and passive versions. While the passive algorithm works better for congestion controls with RTT-independent convergence, it can still produce oscillations on short time scales. The passive algorithm, described in <xreftarget="example-alg-pas"/>,target="example-alg-pas" format="default"/>, is therefore consideredashighly experimental and not safe to deploy outside of testbed environments.Figure 2<xref target="fig_2"/> shows the interaction between flows and theFSE,FSE using the variable names defined in <xreftarget="fse-variables"/>.</t>target="fse-variables" format="default"/>.</t> <figurealign="center">anchor="fig_1" title="Coupled congestion control architecture" > <artworkalign="center"> <![CDATA[ -------align="center" name="" type="" alt=""><![CDATA[------- <--- Flow 1 | FSE | <--- Flow 2 .. ------- <--- .. Flow N ^ | | ------- | | SBD | <-------| -------]]> </artwork> <postamble>Figure 1: Coupled congestion control architecture</postamble> </figure>]]></artwork></figure> <figurealign="center">anchor="fig_2" title="Flow-FSE interactions"> <artworkalign="center"> <![CDATA[ Flow#1(cc)align="center" name="" type="" alt=""><![CDATA[Flow#1(cc) FSE Flow#2(cc) ---------- --- ---------- #1 JOIN ----register--> REGISTER REGISTER <--register-- JOIN #1 #2 CC_R(1) ----UPDATE----> UPDATE (in) #3 NEW RATE <---FSE_R(1)-- UPDATE (out) --FSE_R(2)-> #3 NEW RATE]]> </artwork> <postamble>Figure 2: Flow-FSE interaction</postamble> </figure>]]></artwork></figure> <t>Since everything shown inFigure 1<xref target="fig_1"/> is assumed to operate on a single host (the sender) only, this document only describes aspects that have an influence on the resulting on-the-wire behavior. It does not, for instance, define how many bits must be used to representFGIs,FGIs or in which way the entities communicate.</t> <t>Implementations can take variousforms:forms; for instance, all the elements in the figure could be implemented within a single application, thereby operating on flows generated by that application only. Another alternative could be to implement both the FSE and SBD together in a separate processwhichthat different applications communicate with via some form of Inter-Process Communication (IPC). Such an implementation would extend the scope to flows generated by multiple applications. The FSE and SBD could also be included in the Operating System kernel. However, only one type of coupling algorithm should be used for all flows. Combinations of multiple algorithms at different aggregation levels (e.g., the Operating System coupling application aggregates with one algorithm, and applications coupling their flows with another) have not been tested and are therefore not recommended. </t> </section> <sectiontitle="Roles" anchor='roles'>anchor="roles" numbered="true" toc="default"> <name>Roles</name> <t>This section gives an overview of the roles of the elements of coupled congestioncontrol,control and provides an example of how coupled congestion control can operate.</t> <sectiontitle="SBD">numbered="true" toc="default"> <name>SBD</name> <t>SBD uses knowledge about the flows to determine which flows belong in the same Flow Group(FG),(FG) and assigns FGIs accordingly. This knowledge can be derived in three basic ways:<list style="numbers"> <t>From</t> <ol spacing="normal" type="1"> <li>From multiplexing:itIt can be based on the simple assumption that packets sharing the same five-tuple (IP source and destination address, protocol, andtransport layertransport-layer port number pair) and having the same values for the Differentiated Services Code Point (DSCP) and the ECN field in the IP header are typically treated in the same way along the path. This method is the only one specified in thisdocument:document; SBDMAY<bcp14>MAY</bcp14> consider all flows that use the same five-tuple,DSCPDSCP, and ECN field value to belong to the same FG. This classification applies to certaintunnels,tunnels or RTP flows that are multiplexed over one transport (cf. <xreftarget="transport-multiplex"/>).target="TRANSPORT-MULTIPLEX" format="default"/>). Such multiplexing is also a recommended usage of RTP inrtcwebWebRTC <xreftarget="rtcweb-rtp-usage"/>.</t> <t>Viatarget="I-D.ietf-rtcweb-rtp-usage" format="default"/>.</li> <li>Via configuration:e.g.e.g., by assuming that a common wireless uplink is also a sharedbottleneck.</t> <t>Frombottleneck.</li> <li>From measurements:e.g.e.g., by considering correlations among measured delay and loss as an indication of a sharedbottleneck.</t> </list> </t>bottleneck.</li> </ol> <t>The methods above have some essentialtrade-offs: e.g.,trade-offs. For example, multiplexing is a completely reliable measure,howeverbut it is limited in scope to twoend pointsendpoints (i.e., it cannot be applied to couple congestion controllers of one sender talking to multiple receivers). A measurement-based SBD mechanism is described in <xreftarget="RFC8382"/>.target="RFC8382" format="default"/>. Measurements can never be 100% reliable, in particular because they are based on thepastpast, but applying coupled congestion controlmeans to makeinvolves making an assumption about the future; it is therefore recommended to implement cautionary measures,e.g.e.g., by disabling coupled congestion control if enabling it causes a significant increase in delay and/or packet loss. Measurements also take time, which entails a certain delay for turning on coupling (refer to <xreftarget="RFC8382"/>target="RFC8382" format="default"/> for details).Using system configuration to decide about shared bottlenecksWhen this is possible, it can be more efficient(fastertoobtain) than using measurements, but it reliesstatically configure shared bottlenecks (e.g., via a system configuration or user input) based on assumptions about the network environment.</t> </section> <sectiontitle="FSE" anchor="fse-variables">anchor="fse-variables" numbered="true" toc="default"> <name>FSE</name> <t>The FSE contains a list of all flows that have registered with it. For each flow,itthe FSE stores the following:<list style="symbols"> <t>a</t> <ul spacing="normal"> <li>a unique flow number f to identify theflow.</t> <t>theflow.</li> <li>the FGI of the FG that it belongs to (based on the definitions in this document, a flow has only onebottleneck,bottleneck and can therefore be in only oneFG).</t> <t>aFG).</li> <li>a priority P(f), which is apositive number,number greater thanzero.</t> <t>Thezero.</li> <li>The rate used by the flow in bits per second, FSE_R(f).</t> <t>The</li> <li>The desired rate DR(f) of flow f. This can be smaller than FSE_R(f) if the application feeding into the flow has less data to send than FSE_R(f) wouldallow,allow or if a maximum value is imposed on the rate. In the absence of suchlimitslimits, DR(f) must be set to the sending rate provided by the congestion control module of flowf.</t> </list>f.</li> </ul> <t> Note that the absolute range of priorities does notmatter:matter; the algorithm works with a flow's priority portion of the sum of all priority values. For example, if there are two flows, flow 1 with priority 1 and flow 2 with priority 2, the sum of the priorities is 3. Then, flow 1 will be assigned 1/3 of the aggregate sendingraterate, and flow 2 will be assigned 2/3 of the aggregate sending rate. Priorities can be mapped to the "very-low", "low","medium""medium", or "high" priority levels described in <xreftarget="I-D.ietf-rtcweb-transports"/>target="I-D.ietf-rtcweb-transports" format="default"/> by simply using the values 1, 2,44, and 8, respectively. </t> <t>In the FSE, each FG contains one staticvariable S_CRvariable, S_CR, which is the sum of the calculated rates of all flows in the same FG. This value is used to calculate the sending rate. </t> <t>The information listed here is enough to implement the sample flow algorithm given below. FSE implementations could easily be extended to store, e.g., a flow's current sending rate for statistics gathering or future potential optimizations.</t> </section> <sectiontitle="Flows" anchor='flows'>anchor="flows" numbered="true" toc="default"> <name>Flows</name> <t>Flows register themselves with SBD and FSE when they start, deregister from the FSE when they stop, and carry out an UPDATE function call every time their congestion controller calculates a new sending rate. Via UPDATE, they provide the newly calculated rateandand, optionally (if the algorithm supportsit)it), the desired rate. The desired rate is less than the calculated rate in case of application-limited flows; otherwise, it is the same as the calculated rate.</t> <t>Below, two example algorithms are described. While other algorithms could be used instead, the same algorithm must be applied to all flows. Names of variables used in the algorithms are explained below.<list style="symbols"> <t> CC_R(f) - The</t> <dl newline="false" indent="10" spacing="normal"> <dt>CC_R(f)</dt><dd>The rate received from the congestion controller of flow f when it callsUPDATE.</t> <t> FSE_R(f) - TheUPDATE.</dd> <dt>FSE_R(f)</dt><dd>The rate calculated by the FSE for flowf.</t> <t> DR(f) - Thef.</dd> <dt>DR(f)</dt><dd>The desired rate of flowf.</t> <t> S_CR - Thef.</dd> <dt>S_CR</dt><dd>The sum of the calculated rates of all flows in the same FG; this value is used to calculate the sendingrate.</t> <t> FG - Arate.</dd> <dt>FG</dt><dd>A group of flows having the sameFGI,FGI andhencehence, sharing the samebottleneck.</t> <t> P(f) - Thebottleneck.</dd> <dt>P(f)</dt><dd>The priority of flowff, which is received from theflow’sflow's congestion controller; the FSE uses this variable for calculatingFSE_R(f).</t> <t> S_P - TheFSE_R(f).</dd> <dt>S_P</dt><dd>The sum of all thepriorities.</t> <t> TLO - Thepriorities.</dd> <dt>TLO</dt><dd>The total leftoverrate:rate; the sum of rates that could not be assigned to flows that were limited by their desiredrate.</t> <t> AR - Therate.</dd> <dt>AR</dt><dd>The aggregate rate that is assigned to flows that are not limited by their desiredrate.</t> </list> </t>rate.</dd> </dl> <sectiontitle="Example algorithmanchor="example-alg-act" numbered="true" toc="default"> <name>Example Algorithm 1 - ActiveFSE" anchor='example-alg-act'>FSE</name> <t>This algorithm was designed to be the simplest possible method to assign rates according to the priorities of flows.SimulationsSimulation results in <xreftarget="fse"></xref>target="FSE" format="default"/> indicate that it doeshowever notnot, however, significantly reduce queuing delay and packet loss.</t><t><list style="format (%d)"> <t>When<ol spacing="normal" type="(%d)"> <li>When a flow f starts, it registers itself with SBD and the FSE. FSE_R(f) is initialized with the congestion controller's initial rate. SBD will assign the correct FGI. When a flow is assigned an FGI, it adds its FSE_R(f) toS_CR.</t> <t>WhenS_CR.</li> <li>When a flow f stops or pauses, its entry is removed from thelist.</t>list.</li> <li> <t>Every time the congestion controller of the flow f determines a new sending rate CC_R(f), the flow calls UPDATE, which carries out the tasks listed below to derive the new sending rates for all the flows in the FG. A flow's UPDATE function uses three local(i.e.(i.e., per-flow) temporary variables: S_P,TLOTLO, and AR.<list style="format (%c)"></t> <ol spacing="normal" type="(%c)"> <li> <t> It updates S_CR.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ S_CR = S_CR + CC_R(f) -FSE_R(f)]]> </artwork> </figure> </t>FSE_R(f) ]]></sourcecode> </li> <li> <t> It calculates the sum of all the priorities, S_P, and initializes FSE_R.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ S_P = 0 for all flows i in FG do S_P = S_P + P(i) FSE_R(i) = 0 endfor]]> </artwork> </figure> </t>for ]]></sourcecode> </li> <li> <t> It distributes S_CR among all flows, ensuring that each flow's desired rate is not exceeded.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ TLO = S_CR while(TLO-AR>0 and S_P>0) AR = 0 for all flows i in FG do if FSE_R[i] < DR[i] then if TLO * P[i] / S_P >= DR[i] then TLO = TLO - DR[i] FSE_R[i] = DR[i] S_P = S_P - P[i] else FSE_R[i] = TLO * P[i] / S_P AR = AR + TLO * P[i] / S_P end if end if end for endwhile]]> </artwork> </figure> </t>while ]]></sourcecode> </li> <li> <t> It distributes FSE_R to all the flows.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ for all flows i in FG do send FSE_R(i) to the flow i endfor]]> </artwork> </figure> </t> </list></t> </list></t>for ]]></sourcecode> </li> </ol> </li> </ol> </section> <sectiontitle="Example algorithmanchor="example-alg-act-cons" numbered="true" toc="default"> <name>Example Algorithm 2 - Conservative ActiveFSE" anchor='example-alg-act-cons'>FSE</name> <t>This algorithm changes algorithm 1 to conservatively emulate the behavior of a single flow by proportionally reducing the aggregate rate on congestion.SimulationsSimulation results in <xreftarget="fse"></xref>target="FSE" format="default"/> indicate that it can significantly reduce queuing delay and packet loss.<!--It misses some features that we would like to incorporate in future versions of this document (e.g. letting bulk transfers immediately use the bandwidth that is not used by application-limited flows); if these features make the algorithm significantly more complex, this will be included as a third variant of the algorithm.--></t> <t>Step (a)</t> <t>Step (a) of the UPDATE function is changed as described below. This also introduces a local variable DELTA, which is used to calculate the difference between CC_R(f) and the previously stored FSE_R(f). To prevent flows from either ignoring congestion or overreacting, a timer keeps them from changing their rates immediately after the common rate reduction that follows a congestion event. This timer is set to2two RTTs of the flow that experienced congestion because it is assumed that a congestion event can persist for up to one RTT of that flow, with another RTT added to compensate for fluctuations in the measured RTT value.<list style="format (%c)"></t> <ol type="(%c)" spacing="normal"> <li> <t> It updates S_CR based on DELTA.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ if Timer has expired or was not set then DELTA = CC_R(f) - FSE_R(f) if DELTA < 0 then // Reduce S_CR proportionally S_CR = S_CR * CC_R(f) / FSE_R(f) Set Timer for 2 RTTs else S_CR = S_CR + DELTA end if end if]]> </artwork> </figure> </t> </list></t>]]></sourcecode> </li> </ol> </section> </section> </section> <section anchor="Application"title="Application">numbered="true" toc="default"> <name>Application</name> <t>This section specifies how the FSE can be applied to specific congestion control mechanisms and makes general recommendations that facilitate applying the FSE to future congestion controls.</t> <section anchor="app-NADA"title="NADA">numbered="true" toc="default"> <name>NADA</name> <t>Network-Assisted DynamicAdapationAdaptation (NADA) <xreftarget="I-D.ietf-rmcat-nada"></xref>target="RFC8698" format="default"/> is a congestion control scheme forrtcweb.WebRTC. It calculates a reference rate r_ref upon receiving anacknowledgment,acknowledgment and then, based on the reference rate,itcalculates a video target rate r_vin and a sending rate for the flows, r_send.</t> <t>When applying the FSE to NADA, the UPDATE function call described in <xreftarget="flows"></xref>target="flows" format="default"/> gives the FSE NADA's reference rate r_ref. The recommended algorithm for NADA is the Active FSE in <xreftarget="example-alg-act"></xref>.target="example-alg-act" format="default"/>. In step 3(c),(d), when the FSE_R(i) is "sent" to the flow i,this means updating r_ref(r_vinr_ref (r_vin and r_send) of flow i is updated with the value of FSE_R(i).</t><!-- <t>NADA simulation results are available from http://heim.ifi.uio.no/safiquli/coupled-cc/. The next version of this document will refer to a technical report that will be made available at the same URL.</t> --></section> <section anchor="app-general"title="General recommendations">numbered="true" toc="default"> <name>General Recommendations</name> <t>This section provides general advice for applying the FSE to congestion control mechanisms.</t><t><list style="hanging" hangIndent="6"> <t hangText="Receiver-side calculations:"> <vspace /><dl newline="true" spacing="normal" indent="6"> <dt>Receiver-side calculations:</dt> <dd> When receiver-side calculations make assumptions about the rate of the sender, the calculations need to besynchronizedsynchronized, or the receiver needs to be updated accordingly. This applies toTFRCTCP Friendly Rate Control (TFRC) <xreftarget="RFC5348"/>,target="RFC5348" format="default"/>, for example, where simulations showed somewhat less favorable results when using the FSE without a receiver-side change <xreftarget="fse" />.</t> <t hangText="Stateful algorithms:"> <vspace />target="FSE" format="default"/>.</dd> <dt>Stateful algorithms:</dt> <dd> When a congestion control algorithm is stateful (e.g.,TCP, with Slow Start, Congestion Avoidance and Fast Recovery),during the TCP slow start, congestion avoidance, or fast recovery phase), these states should be carefully considered such that the overall state of the aggregate flow is correct. This may require sharing more information in the UPDATE call.</t> <t hangText="Rate jumps:"> <vspace /></dd> <dt>Rate jumps:</dt> <dd> The FSE-based coupling algorithms can let a flow quickly increase its rate to its fair share,e.g.e.g., when a new flow joins or after a quiescent period. In case of window-based congestion controls, this may produce a burstwhichthat should be mitigated in some way. An example of how this could be done without using a timer is presented in <xreftarget="anrw2016"/>,target="ANRW2016" format="default"/>, using TCP as an example.</t> </list></t></dd> </dl> </section> </section> <section anchor="expected-feedback"title="Expected feedbacknumbered="true" toc="default"> <name>Expected Feedback fromexperiments">Experiments</name> <t>The algorithm described in this memo has so far been evaluated using simulations covering all the tests for more than one flow from <xreftarget="I-D.ietf-rmcat-eval-test"/>target="I-D.ietf-rmcat-eval-test" format="default"/> (see <xreftarget="IETF-93"/>,target="IETF-93" format="default"/> and <xreftarget="IETF-94"/>).target="IETF-94" format="default"/>). Experiments should confirm these results using at least the NADA congestion control algorithm with real-life code (e.g., browsers communicating over an emulated network covering the conditions in <xreftarget="I-D.ietf-rmcat-eval-test"/>.target="I-D.ietf-rmcat-eval-test" format="default"/>). The tests with real-life code should be repeated afterwards in real network environments and monitored. Experiments should investigate cases where the media coder's output rate is below the rate that is calculated by the coupling algorithm (FSE_R(i) in algorithms 1 (<xref target="example-alg-act"/>) and2, section 5.3).2 (<xref target="example-alg-act-cons"/>)). Implementers and testers are invited to document their findings in anInternet draft. </t>Internet-Draft.</t> </section> <sectionanchor="Acknowledgements" title="Acknowledgements">anchor="IANA" numbered="true" toc="default"> <name>IANA Considerations</name> <t>This document hasbenefitted from discussions with and feedback from Andreas Petlund, Anna Brunstrom, Colin Perkins, David Hayes, David Ros (who also gave the FSE its name), Ingemar Johansson, Karen Nielsen, Kristian Hiorth, Mirja Kuehlewind, Martin Stiemerling, Spencer Dawkins, Varun Singh, Xiaoqing Zhu, and Zaheduzzaman Sarker. The authors would like to especially thank Xiaoqing Zhu and Stefan Holmer for helping with NADA and GCC, and Anna Brunstrom as well as Julius Flohr for helping us correct the active algorithm for the case of application-limited flows.</t> <t>This work was partially funded by the European Community under its Seventh Framework Programme through the Reducing Internet Transport Latency (RITE) project (ICT-317700).</t> </section> <section anchor="IANA" title="IANA Considerations"> <t>This memo includesnorequest to IANA.</t>IANA actions.</t> </section> <section anchor="Security"title="Security Considerations">numbered="true" toc="default"> <name>Security Considerations</name> <t>In scenarios where the architecture described in this document is applied across applications, various cheating possibilitiesarise:arise, e.g., supporting wrong values for the calculated rate,thedesired rate, orthepriority of a flow. In the worst case, such cheating could either prevent other flows from sending or make them send at a rate that is unreasonably large. The end result would be unfair behavior at the network bottleneck, akin to what could be achieved with anyUDP basedUDP-based application. Hence, since this is no worse than UDP in general, there seems to be no significant harm in using this in the absence of UDP rate limiters.</t> <t>In the case of a single-user system, it should also be in the interest of any application programmer to give the user the best possible experience by using reasonable flow priorities or even letting the user choose them. In a multi-user system, this interest may not be given, and one could imagine the worst case of an "arms race"situation,situation where applications end up setting their priorities to the maximum value. If all applications do this, the end result is a fair allocation in which the priority mechanism is implicitlyeliminated,eliminated and no major harm is done.</t> <t> Implementers should also be aware of the Security Considerations sections of <xreftarget="RFC3124"/>,target="RFC3124" format="default"/>, <xreftarget="RFC5348"/>,target="RFC5348" format="default"/>, and <xreftarget="RFC7478"/>.</t>target="RFC7478" format="default"/>.</t> </section> </middle><!-- *****BACK MATTER ***** --><back><references title="Normative References"> <!--?rfc include="http://xml.resource.org/public/rfc/bibxml/reference.RFC.2119.xml"?--> &RFC2119; &RFC3124; &RFC5348; &I-D.ietf-rmcat-nada; </references> <references title="Informative References"> &RFC7656; &RFC8087; &RFC8382; &I-D.ietf-rmcat-eval-test; &I-D.ietf-rmcat-gcc;<displayreference target="I-D.ietf-rmcat-eval-test" to="RMCAT-PROPOSALS"/> <displayreference target="I-D.ietf-rmcat-gcc" to="GCC-RTCWEB"/> <displayreference target="I-D.ietf-rtcweb-transports" to="WEBRTC-TRANS"/> <displayreference target="I-D.ietf-rtcweb-rtp-usage" to="RTCWEB-RTP-USAGE"/> <displayreference target="I-D.ietf-rtcweb-overview" to="RTCWEB-OVERVIEW"/> <references> <name>References</name> <references> <name>Normative References</name> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.3124.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.5348.xml"/> <referenceanchor="I-D.ietf-rtcweb-transports" target="">anchor="RFC8698" target="https://www.rfc-editor.org/info/rfc8698"> <front><title>Transports<title>Network-Assisted Dynamic Adaptation (NADA): A Unified Congestion Control Scheme forWebRTC</title>Real-Time Media</title> <authorinitials="H." surname="Alvestrand" fullname="H. Alvestrand"></author>initials='X' surname='Zhu' fullname='Xiaoqing Zhu'> <organization/> </author> <author initials='R' surname='Pan' fullname='Rong Pan'> <organization/> </author> <author initials='M' surname='Ramalho' fullname='Michael A. Ramalho'> <organization/> </author> <author initials='S' surname='Mena' fullname='Sergio Mena de la Cruz'> <organization/> </author> <datemonth="October" year="2016"/>month='January' year='2020'/> </front> <seriesInfoname="Internet-draft" value="draft-ietf-rtcweb-transports-17.txt"/>name="RFC" value="8698"/> <seriesInfo name="DOI" value="10.17487/RFC8698"/> </reference>&RFC7478; &I-D.ietf-rtcweb-overview;</references> <references> <name>Informative References</name> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7656.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8087.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.8382.xml"/> <xi:include href="https://xml2rfc.tools.ietf.org/public/rfc/bibxml/reference.RFC.7478.xml"/> <!-- I-D.ietf-rmcat-eval-test: I-D exists --> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-eval-test.xml"/> <!-- I-D.draft-ietf-rmcat-gcc-02: Expired --> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rmcat-gcc.xml"/> <!-- I-D.ietf-rtcweb-transports: I-D exists--> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rtcweb-transports.xml"/> <!-- I-D.ietf-rtcweb-overview: I-D exists --> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rtcweb-overview.xml"/> <!-- I-D.ietf-rtcweb-rtp-usage: I-D exists--> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml3/reference.I-D.ietf-rtcweb-rtp-usage.xml"/> <!-- draft-westerlund-avtcore-transport-multiplexing-07: Expired - unable to use short way--> <referenceanchor="transport-multiplex"anchor='TRANSPORT-MULTIPLEX' target=""> <front> <title>Multiple RTP Sessions on a Single Lower-Layer Transport</title> <authorinitials="M." surname="Westerlund" fullname="M. Westerlund"></author> <author initials="C." surname="Perkins" fullname="C. Perkins"></author> <date month="October" year="2013"/> </front> <seriesInfo name="Internet-draft" value="draft-westerlund-avtcore-transport-multiplexing-07.txt"/> </reference> <reference anchor="rtcweb-rtp-usage" target=""> <front> <title>Web Real-Time Communication (WebRTC): Media Transport and Use of RTP</title> <author initials="C." surname="Perkins" fullname="C. Perkins"></author> <author initials="M." surname="Westerlund" fullname="M. Westerlund"></author>initials='M.' surname='Westerlund' fullname='Magnus Westerlund'> <organization /> </author> <authorinitials="J." surname="Ott" fullname="J. Ott"></author>initials='C.' surname='Perkins' fullname='Colin Perkins'> <organization /> </author> <datemonth="March" year="2016"/>month='October' year='2013' /> </front> <seriesInfoname="Internet-draft" value="draft-ietf-rtcweb-rtp-usage-26.txt"/>name='Internet-Draft' value='draft-westerlund-avtcore-transport-multiplexing-07'/> </reference> <referenceanchor="fse">anchor="FSE" target="http://safiquli.at.ifi.uio.no/paper/fse-tech-report.pdf"> <front> <title>Coupled Congestion Control for RTP Media</title> <authorinitials="S.I."initials="S." surname="Islam" fullname="S.Islam"> </author>Islam"/> <authorinitials="M.W."initials="M." surname="Welzl" fullname="M.Welzl" > </author>Welzl"/> <authorinitials="S.G."initials="S." surname="Gjessing" fullname="SGjessing" > </author>Gjessing"/> <authorinitials="N.K."initials="N." surname="Khademi" fullname="NKhademi" > </author>Khademi"/> <dateyear="2014" />month="March" year="2014"/> </front><seriesInfo name="ACM<refcontent>ACM SIGCOMM Capacity Sharing Workshop (CSWS 2014) and ACM SIGCOMM CCR 44(4)2014; extended version available as a technical report from http://safiquli.at.ifi.uio.no/paper/fse-tech-report.pdf" value="" />2014 </refcontent> </reference> <referenceanchor="fse-noms">anchor="FSE-NOMS"> <front> <title>ManagingReal-Time Media Flowsreal-time media flows through aFlow State Exchange</title>flow state exchange</title> <seriesInfo name="DOI" value="10.1109/NOMS.2016.7502803"/> <authorinitials="S.I."initials="S." surname="Islam"fullname="S. Islam"> </author>fullname="Safiqul Islam"/> <authorinitials="M.W."initials="M." surname="Welzl"fullname="M. Welzl" > </author>fullname="Michael Welzl"/> <authorinitials="D.H"initials="D." surname="Hayes"fullname="D Hayes" > </author>fullname="David Hayes"/> <authorinitials="S.G."initials="S." surname="Gjessing"fullname="S Gjessing" > </author> <date year="2016" />fullname="Stein Gjessing"/> </front><seriesInfo name="IEEE<refcontent>IEEE NOMS2016, Istanbul, Turkey" value="" />2016 </refcontent> </reference> <reference anchor="IETF-93" target="https://www.ietf.org/proceedings/93/rmcat.html"> <front> <title>Updates onCoupled'Coupled Congestion Control for RTPMedia</title>Media' </title> <authorinitials="S.I."initials="S." surname="Islam"fullname="S. Islam"> </author>fullname="Safiqul Islam"/> <authorinitials="M.W."initials="M." surname="Welzl"fullname="M. Welzl" > </author>fullname="Michael Welzl"/> <authorinitials="S.G."initials="S." surname="Gjessing" fullname="SGjessing" > </author>Gjessing"/> <datemonth= "July" year = "2015"/>month="July" year="2015"/> </front> <seriesInfo name="IETF" value="93" /> <refcontent>RTP Media Congestion Avoidance Techniques (rmcat) Working Group</refcontent> </reference> <reference anchor="IETF-94" target="https://www.ietf.org/proceedings/94/rmcat.html"> <front> <title>Updates onCoupled'Coupled Congestion Control for RTPMedia</title>Media'</title> <authorinitials="S.I."initials="S." surname="Islam"fullname="S. Islam"> </author>fullname="Safiqul Islam"/> <authorinitials="M.W."initials="M." surname="Welzl" fullname="M.Welzl" > </author>Welzl"/> <authorinitials="S.G."initials="S." surname="Gjessing" fullname="SGjessing" > </author>Gjessing"/> <datemonth= "November" year = "2015"/>month="November" year="2015"/> </front> <seriesInfo name="IETF" value="94" /> <refcontent>RTP Media Congestion Avoidance Techniques (rmcat) Working Group</refcontent> </reference> <referenceanchor="anrw2016">anchor="ANRW2016"> <front> <title>Start MeUp:DeterminingUp: Determining and SharingTCP’sTCP's Initial Congestion Window</title> <seriesInfo name="Proceedings of the 2016 Applied Networking Research Workshop" value="Pages 52-54"/> <seriesInfo name="DOI" value="10.1145/2959424.2959440"/> <authorinitials="S.I."initials="S." surname="Islam"fullname="S. Islam"> </author>fullname="Safiqul Islam"/> <authorinitials="M.W."initials="M." surname="Welzl"fullname="M. Welzl" > </author>fullname="Michael Welzl"/> <dateyear="2016" />month="July" year="2016"/> </front><seriesInfo name="ACM,<refcontent>ACM, IRTF, ISOC Applied Networking Research Workshop 2016 (ANRW2016)" value="" />2016) </refcontent> </reference> </references> </references> <section anchor="app-GCC"title="Applicationnumbered="true" toc="default"> <name>Application toGCC">GCC</name> <t>Google Congestion Control (GCC) <xreftarget="I-D.ietf-rmcat-gcc"></xref>target="I-D.ietf-rmcat-gcc" format="default"/> is another congestion control scheme for RTP flows that is under development. GCC is not yetfinalised,finalized, but at the time of this writing, the rate control of GCC employs two parts: controlling the bandwidth estimate based ondelay,delay and controlling the bandwidth estimate based on loss. Both are designed to estimate the available bandwidth, A_hat. </t> <t>When applying the FSE to GCC, the UPDATE function call described in <xreftarget="flows"></xref>target="flows" format="default"/> gives the FSE GCC's estimate of available bandwidth A_hat. The recommended algorithm for GCC is the Active FSE in <xreftarget="example-alg-act"></xref>.target="example-alg-act" format="default"/>. In step 3(c),(d) of this algorithm, when the FSE_R(i) is "sent" to the flow i,this means updatingA_hat of flow i is updated with the value of FSE_R(i).</t> </section> <sectiontitle="Scheduling" anchor='scheduling'>anchor="scheduling" numbered="true" toc="default"> <name>Scheduling</name> <t> When flows originate from the same host, it would be possible to use only onesinglesender-side congestion controllerwhichthat determines the overall allowed sendingrate,rate and then use a local scheduler to assign a proportion of this rate to each RTP session. This way, priorities could also be implemented as a function of the scheduler. The Congestion Manager (CM) <xreftarget="RFC3124"/>target="RFC3124" format="default"/> also uses such a scheduling function.</t> </section> <sectiontitle="Example algorithmanchor="example-alg-pas" numbered="true" toc="default"> <name>Example Algorithm - PassiveFSE" anchor='example-alg-pas'>FSE</name> <t>Active algorithms calculate the rates for all the flows in the FG and actively distribute them. In a passive algorithm, UPDATE returns a rate that should be used instead of the rate that the congestion controller has determined. This can make a passive algorithm easier to implement; however, when round-trip times of flows are unequal,shorter-RTTflows with shorter RTTs may (depending on the congestion control algorithm) update and react to the overall FSE state more often thanlonger-RTT flows,flows with longer RTTs, which can produce unwanted side effects. This problem is more significant when the congestion control convergence depends on the RTT. While the passive algorithm works better for congestion controls with RTT-independent convergence, it can still produce oscillations on short time scales. The algorithm described below is therefore consideredashighly experimental and not safe to deploy outside of testbed environments. Results of a simplified passive FSE algorithm with both NADA and GCC can be found in <xreftarget="fse-noms"></xref>.</t>target="FSE-NOMS" format="default"/>.</t> <t>In the passive version of the FSE, TLO(the Total(Total Leftover Rate) is a static variable per FGwhichthat is initialized to 0. Additionally, S_CR is limited to increase or decrease as conservatively as a flow's congestion controller decides in order to prohibit sudden ratejumps.<!--At most, it can be the sum of these calculated rates, as seen by the flow during its last rate update.--></t> <t><list style="format (%d)"> <t>Whenjumps. </t> <ol spacing="normal" type="(%d)"> <li>When a flow f starts, it registers itself with SBD and the FSE. FSE_R(f) and DR(f) are initialized with the congestion controller's initial rate. SBD will assign the correct FGI. When a flow is assigned an FGI, it adds its FSE_R(f) toS_CR.</t> <t>WhenS_CR.</li> <li>When a flow f stops or pauses, it sets its DR(f) to 0 and sets P(f) to-1.</t>-1.</li> <li> <t>Every time the congestion controller of the flow f determines a new sending rate CC_R(f), assuming the flow's new desired rate new_DR(f) to be "infinity" in case of a bulk data transfer with an unknown maximum rate, the flow calls UPDATE, which carries out the tasks listed below to derive the flow's new sending rate, Rate(f). A flow's UPDATE function uses a few local(i.e.(i.e., per-flow) temporary variables, which are all initialized to 0: DELTA,new_S_CRnew_S_CR, and S_P.<list style="format (%c)"></t> <ol spacing="normal" type="(%c)"> <li> <t>For all the flows in its FG (including itself), it calculates the sum of all the calculated rates, new_S_CR.ThenThen, it calculates DELTA: the difference between FSE_R(f) and CC_R(f).<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ for all flows i in FG do new_S_CR = new_S_CR + FSE_R(i) end for DELTA = CC_R(f) -FSE_R(f)]]> </artwork> </figure> </t>FSE_R(f) ]]></sourcecode> </li> <li> <t>It updates S_CR,FSE_R(f)FSE_R(f), and DR(f).<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ FSE_R(f) = CC_R(f) if DELTA > 0 then // the flow's rate has increased S_CR = S_CR + DELTA else if DELTA < 0 then S_CR = new_S_CR + DELTA end if DR(f) =min(new_DR(f),FSE_R(f))]]> </artwork> </figure> </t>min(new_DR(f),FSE_R(f)) ]]></sourcecode> </li> <li> <t>It calculates the leftover rate TLO, removes the terminated flows from theFSEFSE, and calculates the sum of all the priorities, S_P.<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ for all flows i in FG do if P(i)<0 then delete flow else S_P = S_P + P(i) end if end for if DR(f) < FSE_R(f) then TLO = TLO + (P(f)/S_P) * S_CR - DR(f)) endif]]> </artwork> </figure> </t>if ]]></sourcecode> </li> <li> <t>It calculates the sending rate, Rate(f).<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ Rate(f) = min(new_DR(f), (P(f)*S_CR)/S_P + TLO) if Rate(f) != new_DR(f) and TLO > 0 then TLO = 0 // f has 'taken' TLO endif]]> </artwork> </figure> </t>if ]]></sourcecode> </li> <li> <t>It updates DR(f) and FSE_R(f) with Rate(f).<figure align="left"> <artwork align="left"> <![CDATA[</t> <sourcecode type="pseudocode"><![CDATA[ if Rate(f) > DR(f) then DR(f) = Rate(f) end if FSE_R(f) =Rate(f)]]> </artwork> </figure> </t> </list></t> </list></t>Rate(f) ]]></sourcecode> </li> </ol> </li> </ol> <t>The goals of the flow algorithm are to achieve prioritization, improve network utilization in the face of application-limited flows, and impose limits on the increase behavior such that the negative impact of multiple flows trying to increase their rate together is minimized. It does that by assigning a flow a sending rate that may not be what the flow's congestion controller expected. It therefore builds on the assumption that no significant inefficiencies arise from temporary application-limited behavior or from quickly jumping to a rate that is higher than the congestion controller intended. How problematic these issues really are depends on the controllers in use and requires careful per-controller experimentation. The coupled congestion control mechanism described here also does not require all controllers to be equal; effects of heterogeneous controllers, or homogeneous controllers being in different states, are also subject to experimentation.</t> <t>This algorithm givesallthe leftover rate of application-limited flows to the first flow that updates its sending rate, provided that this flow needs it all (otherwise, its own leftover rate can be taken by the next flow that updates its rate). Other policies could be applied,e.g.e.g., to divide the leftover rate of a flow equally among all other flows in the FGI.</t> <section anchor="example-op"title="Example operation (passive)">numbered="true" toc="default"> <name>Example Operation (Passive)</name> <t>In order to illustrate the operation of the passive coupled congestion control algorithm, this section presents a toy example of two flows that use it. Let us assume that both flows traverse a common 10 Mbit/s bottleneck and use a simplistic congestion controller that starts out with 1 Mbit/s, increases its rate by 1 Mbit/s in the absence ofcongestioncongestion, and decreases it by 2 Mbit/s in the presence of congestion. For simplicity, flows are assumed to always operate in a round-robin fashion. Rate numbers below without units are assumed to be in Mbit/s. For illustration purposes, the actual sending rate is also shown for every flow in FSE diagrams even though it is not really stored in the FSE.</t> <t>Flow #1 begins. It is a bulk data transfer and considers itself to have top priority. This is the FSE after the flow algorithm's step 1:</t><figure align="left"><artworkalign="left"> <![CDATA[ ----------------------------------------align="left" name="" type="" alt=""><![CDATA[---------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 1 | 1 | 1 | ---------------------------------------- S_CR = 1, TLO = 0]]> </artwork> </figure>]]></artwork> <t>Its congestion controller gradually increases its rate. Eventually, at some point, the FSE should look like this:</t><figure align="left"><artworkalign="left"> <![CDATA[ -----------------------------------------align="left" name="" type="" alt=""><![CDATA[----------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 10 | 10 | 10 | ----------------------------------------- S_CR = 10, TLO = 0]]> </artwork> </figure> <t>Now]]></artwork> <t>Now, another flow joins. It is also a bulk datatransfer,transfer and has a lower priority (0.5):</t><figure align="left"><artworkalign="left"> <![CDATA[ ------------------------------------------align="left" name="" type="" alt=""><![CDATA[------------------------------------------ | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 10 | 10 | 10 | | 2 | 1 | 0.5 | 1 | 1 | 1 | ------------------------------------------ S_CR = 11, TLO = 0]]> </artwork> </figure> <t>Now]]></artwork> <t>Now, assume that the first flow updates its rate to 8, because the total sending rate of 11 exceeds the total capacity. Let us take a closer look at what happens in step 3 of the flow algorithm.</t><figure align="left"> <artwork align="left"> <![CDATA[ CC_R(1)<t>CC_R(1) = 8. new_DR(1) =infinity. 3 a) new_S_CRinfinity.</t> <ol spacing="normal" type="(3%c)"> <li>new_S_CR = 11; DELTA = 8 - 10 =-2. 3 b) FSE_R(1)-2.</li> <li>FSE_R(1) = 8. DELTA is negative, hence S_CR = 9; DR(1) =8. 3 c) S_P8</li> <li>S_P =1.5. 3 d) new1.5.</li> <li>new sending rate Rate(1) = min(infinity, 1/1.5 * 9 + 0) =6. 3 e) FSE_R(1)6.</li> <li>FSE_R(1) =6. The6.</li> </ol> <t>The resulting FSE looks asfollows:follows:</t> <artwork align="left" name="" type="" alt=""><![CDATA[ ------------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 6 | 8 | 6 | | 2 | 1 | 0.5 | 1 | 1 | 1 | ------------------------------------------- S_CR = 9, TLO = 0]]> </artwork> </figure>]]></artwork> <t>The effect is that flow #1 is sending with 6 Mbit/s instead of the 8 Mbit/s that the congestion controller derived. Let us now assume that flow #2 updates its rate. Its congestion controller detects that the network is not fully saturated (the actual total sending rate is 6+1=7) and increases its rate.</t><figure align="left"> <artwork align="left"> <![CDATA[ CC_R(2)<t>CC_R(2) = 2. new_DR(2) =infinity. 3 a) new_S_CRinfinity.</t> <ol spacing="normal" type="(3%c)"> <li>new_S_CR = 7; DELTA = 2 - 1 =1. 3 b) FSE_R(2)1.</li> <li>FSE_R(2) = 2. DELTA is positive, hence S_CR = 9 + 1 = 10; DR(2) =2. 3 c) S_P2.</li> <li>S_P =1.5. 3 d) Rate(2)1.5.</li> <li>Rate(2) = min(infinity, 0.5/1.5 * 10 + 0) =3.33. 3 e) DR(2)3.33.</li> <li>DR(2) = FSE_R(2) =3.33. The3.33.</li> </ol> <t>The resulting FSE looks asfollows:follows:</t> <artwork align="left" name="" type="" alt=""><![CDATA[ ------------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 6 | 8 | 6 | | 2 | 1 | 0.5 | 3.33 | 3.33 | 3.33 | ------------------------------------------- S_CR = 10, TLO = 0]]> </artwork> </figure>]]></artwork> <t>The effect is that flow #2 is now sending with 3.33 Mbit/s, which is close to half of the rate of flow #1 and leads to a total utilization of 6(#1) + 3.33(#2) = 9.33 Mbit/s. Flow #2's congestion controller has increased its rate faster than the controller actually expected. Now, flow #1 updates its rate. Its congestion controller detects that the network is not fully saturated and increases its rate. Additionally, the application feeding into flow #1 limits the flow's sending rate to at most 2 Mbit/s.</t><figure align="left"> <artwork align="left"> <![CDATA[ CC_R(1)<t>CC_R(1) = 7. new_DR(1) =2. 3 a) new_S_CR2.</t> <ol spacing="normal" type="(3%c)"> <li>new_S_CR = 9.33; DELTA =1. 3 b) FSE_R(1)1.</li> <li>FSE_R(1) = 7, DELTA is positive, hence S_CR = 10 + 1 = 11; DR(1) = min(2, 7) = 2.3 c) S_P</li> <li>S_P = 1.5; DR(1)<< FSE_R(1), hence TLO = 1/1.5 * 11 - 2 =5.33. 3 d) Rate(1)5.33.</li> <li>Rate(1) = min(2, 1/1.5 * 11 + 5.33) =2. 3 e) FSE_R(1)2.</li> <li>FSE_R(1) =2. The2.</li> </ol> <t>The resulting FSE looks asfollows:follows:</t> <artwork align="left" name="" type="" alt=""><![CDATA[ ------------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 2 | 2 | 2 | | 2 | 1 | 0.5 | 3.33 | 3.33 | 3.33 | ------------------------------------------- S_CR = 11, TLO = 5.33]]> </artwork> </figure>]]></artwork> <t>Now, the total rate of the two flows is 2 + 3.33 = 5.33 Mbit/s,i.e.i.e., the network is significantly underutilized due to the limitation of flow #1. Flow #2 updates its rate. Its congestion controller detects that the network is not fully saturated and increases its rate.</t><figure align="left"> <artwork align="left"> <![CDATA[ CC_R(2)<t>CC_R(2) = 4.33. new_DR(2) =infinity. 3 a) new_S_CRinfinity.</t> <ol spacing="normal" type="(3%c)"> <li>new_S_CR = 5.33; DELTA =1. 3 b) FSE_R(2)1.</li> <li>FSE_R(2) = 4.33. DELTA is positive, hence S_CR = 12; DR(2) =4.33. 3 c) S_P4.33.</li> <li>S_P =1.5. 3 d) Rate(2)1.5.</li> <li>Rate(2) = min(infinity, 0.5/1.5 * 12 + 5.33 ) =9.33. 3 e) FSE_R(2)9.33.</li> <li>FSE_R(2) = 9.33, DR(2) =9.33. The9.33.</li> </ol> <t>The resulting FSE looks asfollows:follows:</t> <artwork align="left" name="" type="" alt=""><![CDATA[ ------------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 1 | 1 | 1 | 2 | 2 | 2 | | 2 | 1 | 0.5 | 9.33 | 9.33 | 9.33 | ------------------------------------------- S_CR = 12, TLO = 0]]> </artwork> </figure>]]></artwork> <t>Now, the total rate of the two flows is 2 + 9.33 = 11.33 Mbit/s. Finally, flow #1 terminates. It sets P(1) to -1 and DR(1) to 0. Let us assume that it terminated late enough for flow #2 to still experience the network in a congested state,i.e.i.e., flow #2 decreases its rate in the next iteration.</t><figure align="left"> <artwork align="left"> <![CDATA[ CC_R(2)<t>CC_R(2) = 7.33. new_DR(2) =infinity. 3 a) new_S_CRinfinity.</t> <ol spacing="normal" type="(3%c)"> <li>new_S_CR = 11.33; DELTA =-2. 3 b) FSE_R(2)-2.</li> <li>FSE_R(2) = 7.33. DELTA is negative, hence S_CR = 9.33; DR(2) =7.33. 3 c) Flow7.33.</li> <li>Flow 1 has P(1) = -1, hence it is deleted from the FSE. S_P =0.5. 3 d) Rate(2)0.5.</li> <li>Rate(2) = min(infinity, 0.5/0.5*9.33 + 0) =9.33. 3 e) FSE_R(2)9.33.</li> <li>FSE_R(2) = DR(2) =9.33. The9.33.</li> </ol> <t>The resulting FSE looks asfollows:follows:</t> <artwork align="left" name="" type="" alt=""><![CDATA[ ------------------------------------------- | # | FGI | P | FSE_R | DR | Rate | | | | | | | | | 2 | 1 | 0.5 | 9.33 | 9.33 | 9.33 | ------------------------------------------- S_CR = 9.33, TLO = 0]]> </artwork> </figure>]]></artwork> </section> </section> <sectiontitle="Change log"> <section title="draft-welzl-rmcat-coupled-cc"> <section title="Changes from -00 to -01"> <t> <list style="symbols"> <t> Added change log. </t> <t> Updated the example algorithm and its operation.</t> </list> </t> </section> <section title="Changesanchor="Acknowledgements" numbered="false" toc="default"> <name>Acknowledgements</name> <t>This document benefited from-01 to -02"> <t> <list style="symbols"> <t>Included an active version of the algorithm which is simpler.</t> <t>Replaced "greedy flow" with "bulk data transfer" and "non-greedy"discussions with"application-limited".</t> <t>Updated new_CR to CC_R, and CR to FSE_R for better understanding. </t> </list> </t> </section> <section title="Changes from -02 to -03"> <t> <list style="symbols"> <t>Included an active conservative version of the algorithm which reduces queue growthandpacket loss; added a reference to a technical report that shows these benefits with simulations.</t> <t>Moved the passive variant of the algorithm to appendix.</t> </list> </t> </section> <section title="Changes from -03 to -04"> <t> <list style="symbols"> <t>Extended SBD section.</t> <t>Added a note about window-based controllers.</t> </list> </t> </section> <section title="Changesfeedback from-04 to -05"> <t> <list style="symbols"> <t>Added a section about applying<contact fullname="Andreas Petlund"/>, <contact fullname="Anna Brunstrom"/>, <contact fullname="Colin Perkins"/>, <contact fullname="David Hayes"/>, <contact fullname="David Ros"/> (who also gave the FSEto specific congestion control algorithms, with a subsection specifyingitsuse with NADA.</t> </list> </t> </section> </section> <section title="draft-ietf-rmcat-coupled-cc"> <section title="Changes from draft-welzl-rmcat-coupled-cc-05"> <t> <list style="symbols"> <t>Moved scheduling section to the appendix.</t> </list> </t> </section> <section title="Changes from -00 to -01"> <t> <list style="symbols"> <t>Included how to apply the algorithm to GCC.</t> <t>Updated variable names of NADAname), <contact fullname="Ingemar Johansson"/>, <contact fullname="Karen Nielsen"/>, <contact fullname="Kristian Hiorth"/>, <contact fullname="Mirja Kuehlewind"/>, <contact fullname="Martin Stiemerling"/>, <contact fullname="Spencer Dawkins"/>, <contact fullname="Varun Singh"/>, <contact fullname="Xiaoqing Zhu"/>, and <contact fullname="Zaheduzzaman Sarker"/>. The authors would like tobe in lineespecially thank <contact fullname="Xiaoqing Zhu"/> and <contact fullname="Stefan Holmer"/> for helping withthe latest version.</t> <t>Added a reference to <xref target="I-D.ietf-rtcweb-transports"/> to make a connection to the prioritization text there.</t> </list> </t> </section> <section title="Changes from -01 to -02"> <t> <list style="symbols"> <t>Minor changes.</t> <t>Moved references ofNADA andGCC from informative to normative. </t> <t>Added a reference for the passive variant of the algorithm.</t> </list> </t> </section> <section title="Changes from -02 to -03"> <t> <list style="symbols"> <t>Minor changes.</t> <t>Added a section about expected feedback from experiments.</t> </list> </t> </section> <section title="Changes from -03 to -04"> <t> <list style="symbols"> <t> Described the names of variables used in the algorithms.</t> <t> Added a diagram to illustrate the interaction between flowsGCC, and <contact fullname="Anna Brunstrom"/> as well as <contact fullname="Julius Flohr"/> for helping us correct theFSE. </t> <t> Added text on the trade-off of using the configuration based approach.</t> <t> Minor changes to enhanceactive algorithm for thereadability.</t> </list> </t> </section> <section title="Changes from -04 to -05"> <t> <list style="symbols"> <t> Changed several occurrencescase of"NADA and GCC" to "NADA", including the abstract.</t> <t> Moved the application to GCC to an appendix, and made the GCC reference informative.</t> <t> Provided a few more general recommendations on applying the coupling algorithm.</t> </list> </t> </section> <section title="Changes from -05 to -06"> <t> <list style="symbols"> <t> Incorporated commentsapplication-limited flows.</t> <t>This work was partially funded byColin Perkins.</t> </list> </t> </section> <section title="Changes from -06 to -07"> <t> <list style="symbols"> <t>Addressed OPSDIR, SECDIR, GENART, AD and IESG comments.</t> </list> </t> </section> <section title="Changes from -07 to -08"> <t> <list style="symbols"> <t>Updatedthealgorithms in section 5 to support application-limited flows. Moved definition of Desired Rate from appendix to section 5. Updated references.</t> </list> </t> </section> <section title="Changes from -08 to -09"> <t> <list style="symbols"> <t>Minor improvement ofEuropean Community under its Seventh Framework Program through thealgorithms in section 5.</t> </list> </t> </section> </section>Reducing Internet Transport Latency (RITE) project (ICT-317700).</t> </section> </back> </rfc>