]> Fastly

jri.ietf@gmail.com

Google

ianswett@google.com

Ericsson

mirja.kuehlewind@ericsson.com

Transport QUIC This document describes a QUIC extension for an endpoint to control its peer's delaying of acknowledgements. Note to Readers Discussion of this draft takes place on the QUIC working group mailing list (quic@ietf.org), which is archived at . Source code and issues list for this draft can be found at . Working Group information can be found at .

Introduction This document describes a QUIC extension for an endpoint to control its peer's delaying of acknowledgements.

Terms and Definitions The keywords "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here. In the rest of this document, "sender" refers to a QUIC data sender (and acknowledgement receiver). Similarly, "receiver" refers to a QUIC data receiver (and acknowledgement sender). An "acknowledgement packet" refers to a QUIC packet that contains only an ACK frame. This document uses terms, definitions, and notational conventions described in Section and Section of .

Motivation A receiver acknowledges received packets, but it can delay sending these acknowledgements. Delaying acknowledgements can impact connection throughput, loss detection and congestion controller performance at a data sender, and CPU utilization at both a data sender and a data receiver. Reducing the frequency of acknowledgements can improve connection and endpoint performance in the following ways:

• Sending UDP packets can be very CPU intensive on some platforms. Reducing the number of packets that only contain acknowledgements reduces the CPU consumed at a data receiver. Experience shows that this reduction can be critical for high bandwidth connections.
• Similarly, receiving and processing UDP packets can also be CPU intensive, and reducing acknowledgement frequency reduces this cost at a data sender.
• For asymmetric link technologies, such as DOCSIS, LTE, and satellite, connection throughput in the forward path can become constrained when the reverse path is filled by acknowledgment packets . When traversing such links, reducing the number of acknowledgments can achieve higher connection throughput, lower the impact on other flows or optimise the overall use of transmission resources .
• The rate of acknowledgment packets can impact link efficiency, including transmission opportunities or battery life, as well as transmission opportunities available to other flows sharing the same link.

As discussed in however, there can be undesirable consequences to congestion control and loss recovery if a receiver uniltaerally reduces the acknowledgment frequency. A sender's constraints on the acknowledgement frequency need to be taken into account to maximize congestion controller and loss recovery performance. specifies a simple delayed acknowledgement mechanism that a receiver can use: send an acknowledgement for every other packet, and for every packet that is received out of order (Section 13.2.1 of ). This does not allow a sender to signal its preferences or constraints. This extension provides a mechanism to solve this problem.

Negotiating Extension Use Endpoints advertise their support of the extension described in this document by sending the following transport parameter ():

min_ack_delay (0xff04de1a):

A variable-length integer representing the minimum amount of time in microseconds by which the endpoint that is sending this value is able to delay an acknowledgement. This limit could be based on the receiver's clock or timer granularity. 'min_ack_delay' is used by the peer to avoid requesting too small a value in the Request Max Ack Delay field of the ACK_FREQUENCY frame.

An endpoint's min_ack_delay MUST NOT be greater than its max_ack_delay. Endpoints that support this extension MUST treat receipt of a min_ack_delay that is greater than the received max_ack_delay as a connection error of type TRANSPORT_PARAMETER_ERROR. Note that while the endpoint's max_ack_delay transport parameter is in milliseconds (), min_ack_delay is specified in microseconds. The min_ack_delay transport parameter is a unilateral indication of support for receiving ACK_FREQUENCY frames. If an endpoint sends the transport parameter, the peer is allowed to send ACK_FREQUENCY frames independent of whether it also sends the min_ack_delay transport parameter or not. Receiving a min_ack_delay transport parameter indicates that the peer might send ACK_FREQUENCY frames in the future. Until an ACK_FREQUENCY frame is received, receiving this transport parameter does not cause the endpoint to change its acknowledgement behavior. Endpoints MUST NOT remember the value of the min_ack_delay transport parameter they received for use in a subsequent connection. Consequently, ACK_FREQUENCY frames cannot be sent in 0-RTT packets, as per . This Transport Parameter is encoded as per .

ACK_FREQUENCY Frame Delaying acknowledgements as much as possible reduces both work done by the endpoints and network load. An endpoint's loss detection and congestion control mechanisms however need to be tolerant of this delay at the peer. An endpoint signals the frequency it wants to receive ACK frames to its peer using an ACK_FREQUENCY frame, shown below: Following the common frame format described in , ACK_FREQUENCY frames have a type of 0xaf, and contain the following fields:

Sequence Number:

A variable-length integer representing the sequence number assigned to the ACK_FREQUENCY frame by the sender to allow receivers to ignore obsolete frames.

Ack-Eliciting Threshold:

A variable-length integer representing the maximum number of ack-eliciting packets the recipient of this frame receives before sending an acknowledgment. A receiving endpoint SHOULD send at least one ACK frame when more than this number of ack-eliciting packets have been received. A value of 0 results in a receiver immediately acknowledging every ack-eliciting packet. By default, an endpoint sends an ACK frame for every other ack-eliciting packet, as specified in , which corresponds to a value of 1.

Request Max Ack Delay:

A variable-length integer representing the value to which the endpoint requests the peer update its max_ack_delay (). The value of this field is in microseconds, unlike the 'max_ack_delay' transport parameter, which is in milliseconds. Sending a value smaller than the min_ack_delay advertised by the peer is invalid. Receipt of an invalid value MUST be treated as a connection error of type PROTOCOL_VIOLATION. On receiving a valid value in this field, the endpoint MUST update its max_ack_delay to the value provided by the peer.

Reordering Threshold:

A variable-length integer that indicates the maximum packet reordering before eliciting an immediate ACK. If no ACK_FREQUENCY frames have been received, the endpoint immediately acknowledges any subsequent packets that are received out of order, as specified in , corresponding to a default value of 1. A value of 0 indicates out-of-order packets do not elicit an immediate ACKs.

ACK_FREQUENCY frames are ack-eliciting. When an ACK_FREQUENCY frame is lost, the sender is encouraged to send another ACK_FREQUENCY frame, unless an ACK_FREQUENCY frame with a larger Sequence Number value has already been sent. However, it is not forbidden to retransmit the lost frame (see Section 13.3 of ), as the receiver will ignore duplicate or out-of-order ACK_FREQUENCY frames based on the Sequence Number. An endpoint can send multiple ACK_FREQUENCY frames with different values within a connection. A sending endpoint MUST send monotonically increasing values in the Sequence Number field, since this field allows ACK_FREQUENCY frames to be processed out of order. A receiving endpoint MUST ignore a received ACK_FREQUENCY frame if the Sequence Number value in the frame is smaller than the largest processed thus far.

IMMEDIATE_ACK Frame A sender can use an ACK_FREQUENCY frame to reduce the number of acknowledgements sent by a receiver, but doing so increases the chances that time-sensitive feedback is delayed as well. For example, as described in , delaying acknowledgements can increase the time it takes for a sender to detect packet loss. The IMMEDIATE_ACK frame helps mitigate this problem. An IMMEDIATE_ACK frame can be useful in other situations as well. For example, if a sender wants an immediate RTT measurement or if a sender wants to establish receiver liveness as quickly as possible. PING frames () are ack-eliciting but if a PING frame is sent without an IMMEDIATE_ACK frame, the receiver might not immediately send an ACK based on its local ACK strategy. By definition IMMEDIATE_ACK frames are ack-eliciting. An endpoint SHOULD send a packet containing an ACK frame immediately upon receiving an IMMEDIATE_ACK frame. An endpoint MAY delay sending an ACK frame despite receiving an IMMEDIATE_ACK frame. For example, an endpoint might do this if a large number of received packets contain an IMMEDIATE_ACK or if the endpoint is under heavy load.

Sending Acknowledgments Prior to receiving an ACK_FREQUENCY frame, endpoints send acknowledgements as specified in . On receiving an ACK_FREQUENCY frame and updating its max_ack_delay and Ack-Eliciting Threshold values (), the endpoint sends an acknowledgement when one of the following conditions are met:

• Since the last acknowledgement was sent, the number of received ack-eliciting packets is greater than the Ack-Eliciting Threshold.
• Since the last acknowledgement was sent, max_ack_delay amount of time has passed.

, , and describe exceptions to this strategy.

Response to Out-of-Order Packets As specified in , endpoints are expected to send an acknowledgement immediately on receiving a reordered ack-eliciting packet. This extension modifies that behavior when an ACK_FREQUENCY frame with a Reordering Threshold value other than 1 has been received.

Largest Unacked:

The largest packet number among all received ack-eliciting packets.

Largest Acked:

The Largest Acknowledged value sent in an ACK frame.

Unreported Missing:

Packets with packet numbers between the Largest Unacked and Largest Acked that have not yet been received.

An endpoint that receives an ACK_FREQUENCY frame with a non-zero Reordering Threshold value SHOULD send an immediate ACK when the gap between the smallest Unreported Missing packet and the Largest Unacked is greater than or equal to the Reordering Threshold value. Sending this additional ACK will reset the max_ack_delay timer and Ack-Eliciting Threshold counter as any ACK would do. In order to ensure timely loss detection, it is optimal to send a Reordering Threshold value of 1 less than the packet threshold used by the data sender for loss detection. If the threshold is smaller, an ACK_FRAME is sent before the packet can be declared lost based on the packet threshold. If the value is larger, it causes unnecessary delays. () recommends a default packet threshold for loss detection of 3, equivalent to a Reordering Threshold of 2. If the most recent ACK_FREQUENCY frame received from the peer has a Reordering Threshold value of 0, the endpoint SHOULD NOT send an immediate acknowledgement in response to packets received out of order, and instead rely on the peer's Ack-Eliciting Threshold and max_ack_delay thresholds for sending acknowledgements.

Examples When the reordering threshold is 1, any time a packet is received and there is a missing packet, an immediate ACK is sent. If the reordering theshold is 3 and ACKs are only sent due to reordering: Receive 1 Receive 3 -> 2 Missing Receive 4 -> 2 Missing Receive 5 -> Send ACK because of 2 Receive 8 -> 6,7 Missing Receive 9 -> Send ACK because of 6, 7 Missing Receive 10 -> Send ACK because of 7 If the reordering threshold is 5 and ACKs are only sent due to reordering: Receive 1 Receive 3 -> 2 Missing Receive 5 -> 2 Missing, 4 Missing Receive 6 -> 2 Missing, 4 Missing Receive 7 -> Send ACK because of 2, 4 Missing Receive 8 -> 4 Missing Receive 9 -> Send ACK because of 4

Expediting Congestion Signals An endpoint SHOULD send an immediate acknowledgement when a packet marked with the ECN Congestion Experienced (CE) codepoint in the IP header is received and the previously received packet was not marked CE. Doing this maintains the peer's response time to congestion events, while also reducing the ACK rate compared to during extreme congestion or when peers are using DCTCP or other congestion controllers (e.g. ) that mark more frequently than classic ECN .

Batch Processing of Packets To avoid sending multiple acknowledgements in rapid succession, an endpoint can process all packets in a batch before determining whether to send an ACK frame in response, as stated in .

Computation of Probe Timeout Period On sending an update to the peer's max_ack_delay, an endpoint can use this new value in later computations of its Probe Timeout (PTO) period; see . Until the frame is acknowledged, the endpoint MUST use the greater of the current max_ack_delay and the value that is in flight when computing the PTO period. Doing so avoids spurious PTOs that can be caused by an update that increases the peer's max_ack_delay. While it is expected that endpoints will have only one ACK_FREQUENCY frame in flight at any given time, this extension does not prohibit having more than one in flight. When using max_ack_delay for PTO computations, endpoints MUST use the maximum of the current value and all those in flight. When the number of in-flight ack-eliciting packets is larger than the ACK-Eliciting Threshold, an endpoint can expect that the peer will not need to wait for its max_ack_delay period before sending an acknowledgement. In such cases, the endpoint MAY therefore exclude the peer's 'max_ack_delay' from its PTO calculation. When Ignore Order is enabled and loss causes the peer to not receive enough packets to trigger an immediate acknowledgement, the receiver will wait 'max_ack_delay', increasing the chances of a premature PTO. Therefore, if Ignore Order is enabled, the PTO MUST be larger than the peer's 'max_ack_delay'.

Determining Acknowledgement Frequency This section provides some guidance on a sender's choice of acknowledgment frequency and discusses some additional considerations. Implementers can select an appropriate strategy to meet the needs of their applications and congestion controllers.

Congestion Control A sender needs to be responsive to notifications of congestion, such as a packet loss or an ECN CE marking. To enable a sender to respond to potential network congestion in a timely fashion, usually at least one acknowledgement per round trip is needed if there are unacknowledged ack-eliciting packets in flight. A sender can accomplish this by setting the Ack-Eliciting Threshold to a value no larger than the current congestion window or the Request Max Ack Delay value to no more than the estimated round trip. Note that the congestion window particularly but also the RTT are dynamic and therefore might require frequent updates if the selected value are close to these limits. Alternatively, a sender can accomplish this by sending an IMMEDIATE_ACK frame once per round trip, though if the packet containing an IMMEDIATE_ACK is lost, detection of that loss will be delayed by the reordering threshold or requested max ack delay. Note that it is possible that the RTT is smaller than the receiver's timer granularity, as communicated via the 'min_ack_delay' transport parameter, preventing the receiver from sending an acknowledgment every RTT in time. In these cases, Reordering Threshold values other than 1 can be harmful, because it delays fast loss detection. A congestion controller that is congestion window limited relies upon receiving acknowledgements to send additional data into the network. An increase in acknowledgement delay increases the delay in sending data, which can reduce the achieved throughput. Congestion window growth can also depend upon receiving acknowledgements. This can be particularly significant in slow start (), when delaying acknowledgements can delay the increase in congestion window and can create larger packet bursts. If the sender is application-limited, acknowledgements can be delayed unnecessarily when entering idle periods. Therefore, if no further data is buffered to be sent, a sender can send an IMMEDIATE_ACK frame with the last data packet before an idle period to avoid waiting for the ack delay.

Burst Mitigation Receiving an acknowledgement can allow a sender to release new packets into the network. If a sender is designed to rely on the timing of peer acknowledgments ("ACK clock"), delaying acknowledgments can cause undesirable bursts of data into the network. In keeping with Section 7.7 of , a sender can either employ pacing or limit bursts to the initial congestion window.

Loss Detection and Timers Acknowledgements are fundamental to reliability in QUIC. Consequently, delaying or reducing the frequency of acknowledgments can cause loss detection at the sender to be delayed. A QUIC sender detects loss using packet thresholds on receiving an acknowledgement (Section 6.1.1 of ); delaying the acknowledgement therefore delays this method of detecting losses. Reducing acknowledgement frequency reduces the number of RTT samples that a sender receives (Section 5 of ), making a sender's RTT estimate less responsive to changes in the path's RTT. As a result, any mechanisms that rely on an accurate RTT estimate, such as time-threshold loss detection (Section 6.1.2 of ) or Probe Timeout (Section 6.2 of ), will be less responsive to changes in the path's RTT, resulting in either delayed or unnecessary packet transmissions. To limit these consequences of reduced acknowledgement frequency, a sender SHOULD cause a receiver to send an acknowledgement at least once per RTT if there are unacknowledged ack-eliciting packets in flight. A sender can accomplish this by sending an IMMEDIATE_ACK frame once per round-trip time (RTT), or it can set the Ack-Eliciting Threshold and Request Max Ack Delay values to be no more than a congestion window and an estimated RTT, respectively. A sender might use timers to detect loss of PMTU probe packets (). A sender SHOULD bundle an IMMEDIATE_ACK frame with any PTMU probes to avoid triggering such timers.

Connection Migration To avoid additional delays to connection migration confirmation when using this extension, a client can bundle an IMMEDIATE_ACK frame with the first non-probing frame () it sends or it can send only an IMMEDIATE_ACK frame, which is a non-probing frame. An endpoint's congestion controller and RTT estimator are reset upon confirmation of migration (), which can impact the number of acknowledgements received after migration. An endpoint that has sent an ACK_FREQUENCY frame earlier in the connection SHOULD update and send a new ACK_FREQUENCY frame immediately upon confirmation of connection migration.

Security Considerations An improperly configured or malicious data sender could cause a data receiver to acknowledge more frequently than its available resources permit. However, there are two limits that make such an attack largely inconsequential. First, the acknowledgement rate is bounded by the rate at which data is received. Second, ACK_FREQUENCY and IMMEDIATE_ACK frames can only request an increase in the acknowledgment rate, but cannot force it. In general, with this extension, a sender cannot force a receiver to acknowledge more frequently than the receiver considers safe based on its resource constraints.

IANA Considerations This document defines a new transport parameter to advertise support of the extension described in this document and two new frame types to registered by IANQ in the respective "QUIC Protocol" registries under https://www.iana.org/assignments/quic/quic.xhtml. The following entry in should be added to the "QUIC Transport Parameters" registry under the "QUIC Protocol" heading. Addition to QUIC Transport Parameters Entries

Value	Parameter Name.	Specification
0xff04de1a	min_ack_delay.

The following frame types should be added to the "QUIC Frame Types" registry under the "QUIC Protocol" heading. Addition to QUIC Frame Types Entries

Value	Frame Name	Specification
0xaf	ACK_FREQUENCY
0xac	IMMEDIATE_ACK

References Normative References Fastly Mozilla Fastly Google In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Informative References University of Aberdeen University of Aberdeen University of Aberdeen This document describes TCP performance problems that arise because of asymmetric effects. These problems arise in several access networks, including bandwidth-asymmetric networks and packet radio subnetworks, for different underlying reasons. However, the end result on TCP performance is the same in both cases: performance often degrades significantly because of imperfection and variability in the ACK feedback from the receiver to the sender. The document details several mitigations to these effects, which have either been proposed or evaluated in the literature, or are currently deployed in networks. These solutions use a combination of local link- layer techniques, subnetwork, and end-to-end mechanisms, consisting of: (i) techniques to manage the channel used for the upstream bottleneck link carrying the ACKs, typically using header compression or reducing the frequency of TCP ACKs, (ii) techniques to handle this reduced ACK frequency to retain the TCP sender's acknowledgment-triggered self- clocking and (iii) techniques to schedule the data and ACK packets in the reverse direction to improve performance in the presence of two-way traffic. Each technique is described, together with known issues, and recommendations for use. A summary of the recommendations is provided at the end of the document. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. This memo specifies the incorporation of ECN (Explicit Congestion Notification) to TCP and IP, including ECN's use of two bits in the IP header. [STANDARDS-TRACK] This Informational RFC describes Data Center TCP (DCTCP): a TCP congestion control scheme for data-center traffic. DCTCP extends the Explicit Congestion Notification (ECN) processing to estimate the fraction of bytes that encounter congestion rather than simply detecting that some congestion has occurred. DCTCP then scales the TCP congestion window based on this estimate. This method achieves high-burst tolerance, low latency, and high throughput with shallow- buffered switches. This memo also discusses deployment issues related to the coexistence of DCTCP and conventional TCP, discusses the lack of a negotiating mechanism between sender and receiver, and presents some possible mitigations. This memo documents DCTCP as currently implemented by several major operating systems. DCTCP, as described in this specification, is applicable to deployments in controlled environments like data centers, but it must not be deployed over the public Internet without additional measures. Nokia Bell Labs Independent CableLabs This specification defines a framework for coupling the Active Queue Management (AQM) algorithms in two queues intended for flows with different responses to congestion. This provides a way for the Internet to transition from the scaling problems of standard TCP-Reno-friendly ('Classic') congestion controls to the family of 'Scalable' congestion controls. These are designed for consistently very low queuing latency, very low congestion loss, and scaling of per-flow throughput by using Explicit Congestion Notification (ECN) in a modified way. Until the Coupled Dual Queue (DualQ), these Scalable L4S congestion controls could only be deployed where a clean-slate environment could be arranged, such as in private data centres. This specification first explains how a Coupled DualQ works. It then gives the normative requirements that are necessary for it to work well. All this is independent of which two AQMs are used, but pseudocode examples of specific AQMs are given in appendices.

Change Log

• RFC Editor's Note: Please remove this section prior to publication of a final version of this document.

Acknowledgments The following people directly contributed key ideas that shaped this draft: Bob Briscoe, Kazuho Oku, Marten Seemann.