]> CUJO AI

Gaustadalléen 21 0349 Norway bjorn.monclair@cujo.com

CUJO AI

Gaustadalléen 21 0349 Norway magnus.olden@cujo.com

CUJO AI

Gaustadalléen 21 0349 Norway ike.kunze@cujo.com

Transport IP Performance Measurement Quality Attenuation Application Outcomes Quality of Outcome Performance monitoring Network quality This document introduces the Quality of Outcome (QoO) network quality score and the corresponding QoO framework as an approach to network quality assessment designed to align with the needs of application developers, users, and operators. By leveraging the Quality Attenuation metric, QoO provides a method for defining and evaluating application-specific, quality-focused network performance requirements to enable insights for network optimization and simple Quality of Service scores for end-users. About This Document Status information for this document may be found at . Discussion of this document takes place on the IP Performance Measurement Working Group mailing list (), which is archived at . Subscribe at . Source for this draft and an issue tracker can be found at .

Introduction This document introduces the Quality of Outcome (QoO) network quality score. It is designed to be easy to understand, while at the same time being objective, adaptable to different network quality needs, and allowing advanced analyses to identify the root cause of network problems (). Centered around the QoO score, this document defines the QoO framework which allows application developers to specify quality-focused network performance requirements (for example, regarding latency, throughput, and packet loss) and provides a way to derive the user-facing QoO score by comparing these application-specific performance requirements to network performance measurements. The QoO framework builds on Quality Attenuation , a network quality metric that enables network operators to achieve fault isolation () and effective network planning through composability . Quality Attenuation meets most of the requirements for a meaningful network quality score set out in : it is composable, captures the ability of a network to satisfy application requirements, and can be compared to a variety of application needs. However, interpreting raw Quality Attenuation values is difficult for end-users and application developers. The challenge is simplifying the entailed information to present results in an understandable and unambiguous way without losing too much precision and accuracy. The QoO framework takes a probabilistic approach to this challenge because network stacks and applications adapt dynamically to changing network conditions. Also, applications and underlying networking protocols make separate optimizations based on the perceived network quality over time, making absolute certainty about outcomes practically impossible. However, educated assessments of expected outcomes remain achievable for which the QoO framework uses a per-application, per application-type, or per-Service Level Agreement (SLA) granularity. This document assumes that network quality can be represented by a minimum required throughput, a set of latency percentiles, and packet loss rates as these measurables will ultimately also capture changes to additional factors. Application developers, regulatory bodies, and other interested parties can use this representation to describe quality-focused network performance requirements. The QoO framework gives structure to this approach by defining two network quality thresholds: one for optimal application performance and one for unacceptable application performance. The QoO score serves as a linear distance measure between the two distinct thresholds and allows network conditions to be expressed in easily understood terms such as "This network provides 94% of optimal conditions for video conferencing (relative to the threshold for unacceptable performance)" while supporting both comprehensive end-to-end tests and analyses from within the network. The QoO framework is designed to be flexible in its application scope. QoO scores may be calculated for the complete end-to-end path (from application client to server), or focused on specific network segments, such as the customer-facing access network, intermediate transit networks, or server-side infrastructure. Through the composability properties of the underlying Quality Attenuation metric, measurements from different segments can be combined or decomposed to isolate performance issues regardless of where they occur in the network path. This document defines a minimum viable framework, and often trades precision for simplicity to facilitate adoption and usability in many different contexts, such as active testing from applications and monitoring from network equipment. Assessing the corresponding loss of precision is important and requires combining measurement results with a description of the measurement approach; provides corresponding guidelines. Note that this document focuses specifically on the general framework of using quality-focused network performance requirements and corresponding network performance measurements to calculate QoO network quality scores. How applications define and share their network performance requirements, which format is used to publish such requirement information, how operators retrieve such data from applications or services, how the precision of the resulting QoO scores is assessed, and what levels of precision are considered acceptable is out of the scope of this document.

Terminology This document uses the following terminology:

Network:

The communication infrastructure that facilitates data transmission between endpoints, including all intermediate devices, links, and protocols that affect the transmission of data. This encompasses both the physical infrastructure and the logical protocols that govern data transmission. The network may support various communication patterns and may span multiple administrative domains.

Network Segment:

A portion of the complete end-to-end network path between application endpoints. A network segment may represent a specific administrative domain (e.g., access network, transit network, or server-side infrastructure), a particular technology domain (e.g., Wi-Fi or cellular), or any subset of the path for which independent quality measurements and analysis are desired.

Quality Attenuation:

A network quality metric defined in that combines latency and packet loss distributions in a unified approach to jointly assess latency and loss characteristics of network performance.

Quality of Experience (QoE):

The degree of delight or annoyance of the user of an application or service. See also .

Quality of Service (QoS):

The totality of characteristics of a telecommunications service that bear on its ability to satisfy stated and implied needs of the user of the service. See also .

Quality of Outcome (QoO):

A network quality framework and metric that evaluates network quality based on how closely measured network conditions meet application-specific, quality-focused performance requirements. QoO is a QoS indicator that may be related to, but cannot be considered the same as, the actual QoE of end-users.

QoO Score:

A numerical value that represents the distance-based assessment of network quality relative to network performance requirements for optimal and unacceptable application performance on a given network for a specific application, typically expressed as a percentage.

Requirements for Optimal Performance (ROP):

The network performance characteristics at which an application achieves optimal performance and quality, beyond which further improvements in network conditions do not result in perceptible improvements in application performance or user experience. When network performance exceeds ROP thresholds, any sub-optimal user experience can be assumed not to be caused by the part of the network path that has been measured for QoO calculations.

Conditions at the Point of Unacceptable Performance (CPUP):

The network performance threshold below which an application fails to provide acceptable user experience. Note that 'unacceptable' in this context refers to degraded performance quality rather than complete technical failure of the application. There is no universally strict threshold defining when network conditions become unacceptable for applications.

Composability:

The mathematical property that allows network quality measurements to be combined across different network segments or decomposed to isolate specific network components for analysis and troubleshooting.

Accuracy and Precision:

"Accuracy" refers to how close measurements are to the value that reflects the real conditions. "Precision" refers to the consistency and repeatability of measurements. These terms are used with their standard statistical meanings and are not interchangeable .

Overview The QoO framework produces simple percentage scores that express network quality in relation to pre-defined, application-specific network performance requirements. For example: "This network provides 94% of optimal conditions for video conferencing". This way, QoO conveys an intuition for how well an application is expected to perform in the assessed network with higher scores intended to convey that applications are more likely to have optimal performance. The QoO framework builds upon Quality Attenuation and evaluates network conditions using latency distributions, packet loss rates, and throughput rates. These measured conditions are compared against application-specific, quality-focused network performance requirements along multiple dimensions (such as 90th or 95th latency percentiles or mean packet loss rate) at two thresholds:

• Optimal performance (ROP): Network conditions under which application performance becomes optimal
• Unacceptable performance (CPUP): Network conditions under which application performance becomes unacceptable

When the measured network conditions fall between the defined thresholds for any of the assessed performance dimensions, the QoO framework calculates a score for each dimension by expressing the current network quality as a relative position (percentage) on the linear scale between the corresponding ROP and CPUP thresholds. The minimum score across all dimensions serves as the overall QoO score for the assessed network based on the rationale that the most degraded performance dimension is likely to determine the application's perceived quality. The dual-threshold approach of the QoO framework enables meaningful scores even when networks are not perfect while accounting for different application sensitivities. The underlying Quality Attenuation measurements are mathematically composable, enabling network operators to isolate performance bottlenecks across different network segments for precise fault diagnosis and network planning. QoO scores are expected to correlate with QoE metrics, such as Mean Opinion Score (MOS) , but they are not designed to deliver MOS or QoE scores directly. QoO measures network service quality, not subjective user experience. It is important to note that the QoO framework itself does not define where QoO scores fall on the spectrum between QoS and QoE metrics. The position on this spectrum depends primarily on how the ROP and CPUP thresholds are chosen. With appropriate threshold selection based on user-acceptance testing and application performance analysis, QoO scores can likely be tuned to be close to (if not identical to) QoE metrics, while still maintaining the objectivity and composability benefits of QoS metrics. The remainder of this document outlines requirements for the QoO framework (), defines the core QoO framework (), discusses operational considerations () and known weaknesses and open questions (), and compares QoO with existing quality metrics ().

Requirements This section describes the features and attributes that a network quality framework must have to be useful for different stakeholders: application developers, end-users, and network operators. At a high level, end-users need an understandable network quality metric. Application developers require a network quality metric that allows them to evaluate how well their application is likely to perform given the measured network performance. Network operators need a metric that facilitates troubleshooting and optimization of their networks. Existing network quality metrics and frameworks address the needs of one or two of these stakeholders, but there is none that bridges the needs of all three. Examples include throughput metrics that operators use to prove regulatory compliance but with little relevance to application performance, or subjective QoE metrics that are understandable to users but difficult for operators to collect at meaningful scale. A key motivation for the QoO framework is to bridge the gap between the technical aspects of network performance and the practical needs of those who depend on it. For example, while solutions exist for many of the problems causing high and unstable latency in the Internet, such as bufferbloat mitigation techniques and improved congestion control algorithms , the incentives to deploy them have remained relatively weak. A unifying framework for assessing network quality can serve to strengthen these incentives significantly. Bandwidth alone is necessary but not sufficient for high-quality modern network experiences. High idle and working latencies, large delay variations, and unmitigated packet loss are major causes of poor application outcomes. The impact of latency is widely recognized in network engineering circles , but benchmarking the quality of network transport remains complex. Most end-users are unable to relate to metrics other than Mbps, which they have long been conditioned to think of as the only dimension of network quality. Real Time Response under load tests and Responsiveness tests make significant strides toward creating a network quality metric that is intended to be closer to application outcomes than bandwidth alone. The latter, in particular, is successful at being relatively relatable and understandable to end-users. However, as noted in , "Our networks remain unresponsive, not from a lack of technical solutions, but rather a lack of awareness of the problem". This lack of awareness means that some operators might have little incentive to improve network quality beyond increasing bandwidth. For example, despite the availability of open-source solutions such as FQ_CoDel , which has been available for over a decade, vendors rarely implement them in widely deployed equipment (e.g., Wi-Fi routers still commonly exhibit bufferbloat). A universally accepted network quality framework that successfully captures the degree to which networks provide the quality required by applications may help to increase the willingness of vendors to implement such solutions. The IAB workshop on measuring Internet quality for end-users identified a key insight: users care primarily about application performance rather than network performance. Among the conclusions was the statement, "A really meaningful metric for users is whether their application will work properly or fail because of a lack of a network with sufficient characteristics" . Therefore, one critical requirement for a meaningful framework is its ability to answer the following question: "Will networking conditions prevent an application from working as intended?". Answering this question requires several considerations. First, the Internet is inherently stochastic from the perspective of any given client, so absolute certainty is unattainable. Second, different applications have different needs and adapt differently to network conditions. A framework aiming to answer the stated question must accommodate such diverse application requirements. Third, end-users have individual tolerances for degradation in network conditions and the resulting effects on application experience. These variations must be factored into the design of a suitable network quality framework.

General Requirements This section describes the requirements for an objective network quality framework and metric that is useful for end-users, application developers, and network operators/vendors alike. Specifically, this section outlines the three main general requirements for such a framework while the sections therafter describe requirements from the perspective of each of the target groups: end-users (), application developers (), and network operators (). In general, all stakeholders ultimately care about the performance of applications running over a network. Application performance does not only depend on bandwidth but also on the delay and delay variation of network links and computational steps involved in making the application function. These delays depend on how the application places load on the network, how the network is affected by environmental conditions, and the behavior of other users and applications sharing the network resources. Likewise, packet loss (e.g., caused by congestion) can also negatively impact application performance in different ways depending on the class of application. Different applications may have different needs from a network and may impose different patterns of load. To determine whether an application will likely work well or fail, a network quality framework must compare measurements of network performance to a wide variety of application requirements. It is important that these measurements reflect the actual network service configuration that will handle the application flows, including any traffic prioritization, network slicing, VPN services, or other differentiated service mechanisms (see ). Flexibility in describing application requirements and the ability to capture the delay and loss characteristics of a network with sufficient accuracy and precision are necessary to compute a meaningful QoO network quality score that can be used to better estimate application performance. The framework must also support spatial composition , to enable operators to take actions when measurements show that applications fail too often. In particular, spatial composition allows results to be divided into sub-results, each measuring the performance of a required sub-milestone that must be reached in time for the application to perform adequately. To summarize, the QoO framework and the corresponding QoO score should have the following properties in order to be meaningful:

1. Capture a set of network performance metrics which provably correlate to the application performance of a set of different applications as perceived by users.
2. Compare meaningfully to different application requirements.
3. Be composable so operators can isolate and quantify the contributions of different sub-outcomes and sub-paths of the network.

Next, the document presents requirements from the perspective of each of the three target groups: end-users (), application developers (), and network operators ().

Requirements from End-Users The QoO framework should facilitate a metric that is based on objective QoS measurements (such as throughput , packet loss , delays , and (one-way) delay variations ), correlated to application performance, and relatively understandable for end-users, similar to QoE metrics, such as MOS . If these requirements are met, QoO is a middle ground between QoS and QoE metrics and allows end-users to understand if a network is a likely source of impairment for what they care about: the outcomes of applications. Examples are how quickly a web page loads, the smoothness of a video conference, or whether or not a video game has any perceptible lag. End-users may have individual tolerances of session quality (i.e., the quality experienced during a single application usage period, such as a video call or gaming session), below which their quality of experience becomes personally unacceptable. However, it may not be feasible to capture and represent these tolerances per user as the user group scales. A compromise is for the QoO framework to place the responsibility for sourcing and representing end-user requirements onto the application developer. Application developers are expected to perform user-acceptance testing (UAT) of their application across a range of users, terminals, and network conditions to determine the terminal and network requirements that will meet the acceptability thresholds for a representative subset of their end-users. Performing UAT helps developers estimate what QoE new end-users are likely to experience based on the application's network performance requirements. These requirements can evolve and improve based on feedback from end-users, and in turn better inform the application's requirements towards the network. Some real world examples where 'acceptable levels' have been derived by application developers include:

• Remote music collaboration: <20ms one-way latency
• Cloud gaming: >15Mbps downlink throughput and 80ms round-trip time (RTT) (specific requirements vary by game and platform; see for an example study on the impact of latency on Counter Strike: Global Offensive)
• Virtual reality (VR): <20ms RTT from head motion to rendered update in VR (; see for latency measurement and interactivity scoring)

Note that developers of similar applications may have arrived at different figures.

Requirements from Application and Platform Developers The QoO framework needs to provide developers the ability to describe the quality-focused network performance requirements of their applications. The network performance requirements must include all relevant dimensions of network quality so that applications sensitive to different network quality dimensions can all evaluate the network accurately. Not all developers have network expertise, so to make it easy for developers to use the framework, developers must be able to specify network performance requirements approximately. Therefore, it must be possible to describe both simple and complex network performance requirements. The framework also needs to be flexible so that it can be used with different kinds of traffic and that extreme network performance requirements which far exceed the needs of today's applications can also be articulated. If these requirements are met, developers of applications and platforms can state or test their network requirements and evaluate whether the network is sufficient for an optimal application outcome. Both the application developers with networking expertise and those without can use the framework.

Requirements from Network Operators and Network Solution Vendors From an operator perspective, the key is to have a framework that allows finding the network quality bottlenecks and objectively comparing different networks, network configurations, and technologies. To achieve this goal, the framework must support mathematically sound compositionality ('addition' and 'subtraction') as network operators rarely manage network traffic end-to-end. If a test is purely end-to-end, the ability to find bottlenecks may be gone. If, however, measurements can be taken in both end-to-end (e.g., a-b-c-d-e) and partial (e.g., a-b-c) fashion, the results can be decomposed to isolate the areas outside the influence of a network operator. In other words, the network quality of a-b-c and d-e can be separated. Compositionality is essential for fault detection and accountability. By having mathematically correct composition, a network operator can measure two segments separately, perhaps even with different approaches, and combine or correlate the results to understand the end-to-end network quality. Another example where composition is useful is troubleshooting a typical web page load sequence over TCP. If web page load times are too slow, DNS resolution time, TCP RTT, and the time it takes to establish TLS connections can be measured separately to get a better idea of where the problem is. A network quality framework should support this kind of analysis to be maximally useful for operators. The framework must be applicable in both lab testing and monitoring of production networks. It must be useful on different time scales, and it cannot have a dependency on network technology or OSI layers. If these requirements are met, network operators can monitor and test their network and understand where the true bottlenecks are, regardless of network technology.

The QoO Framework The QoO framework builds upon the conceptual foundation of the Quality Attenuation metric as proposed in the Broadband Forum standard called Quality of Experience Delivered (QED) . Quality Attenuation represents quality measurements as distributions. Using latency distributions to measure network quality has been proposed by various researchers and practitioners (e.g., , , and ). Quality Attenuation also incorporates a packet loss distribution, treating packet loss as infinite (or too late to be of use, e.g., > 5 seconds) latency , similar to the One-Way Loss Metric for IP Performance Metrics (IPPM) , which defines packet loss as packets that fail to arrive within a specified time threshold. The novelty of Quality Attenuation lies in its unified treatment of latency and loss within a single distributional framework, enabling mathematical composition of network segments as two distributions can be composed using convolution . In line with Quality Attenuation, QoO focuses on latency distributions and packet loss rates as the basis for network quality assessment, with the assumption that additional factors which could in principle be measured but are not explicitly captured will inevitably influence the observed latency and packet loss behavior, so that QoO indirectly accounts for their effects through the measured distributions. The results of corresponding network performance measurements are compared to pre-defined, application-specific network performance requirements, which ultimately yield per-application QoO scores. These scores express network quality as a relative position (percentage) on a linear scale between two performance thresholds: one for optimal performance and one for unacceptable performance. The following sections first discuss how network conditions can be measured (), then describe how QoO defines per-application network performance requirements (), and finally explain how QoO scores are calculated () with an example provided in

Measuring Network Conditions One of the key motivations for the QoO framework is that different applications and application classes fail under different network conditions. For example, downloads are generally more tolerant of latency than real-time applications. Video conferences are often sensitive to high 90th percentile latency and to the difference between the 90th and 99th percentiles. Online gaming typically has a low tolerance for high 99th percentile latency. Similar considerations apply for a variety of other factors with applications usually requiring some minimum level of throughput and tolerating some maximum packet loss rate. Measurement techniques used for QoO must be able to capture these characteristics with sufficient resolution to respect the data granularities needed by the different application classes. Latency distributions can be gathered via both passive monitoring and active testing . The active testing can use any type of traffic, such as connection-oriented TCP and QUIC or connectionless UDP. It can be applied across different layers of the protocol stack and is network technology independent, meaning it can be gathered in an end-user application, within some network equipment, or anywhere in between. Passive methods rely on observing and time-stamping packets traversing the network. Examples of this include TCP SYN and SYN/ACK packets () and the QUIC spin bit . Similar considerations apply to packet loss measurements while throughput measurements usually involve active testing. This document does not mandate the use of specific measurement approaches. However, in addition to measurement approaches standardized in the QED framework , some relevant techniques are:

• Active probing with the Two-Way Active Measurement Protocol (TWAMP) Light , the Simple Two-Way Active Measurement Protocol (STAMP) , or the Isochronous Round-Trip Tester (IRTT)
• Latency Under Load Tests
• Speed Tests with latency measures
• Simulating real traffic
• End-to-end measurements of real traffic
• TCP SYN ACK or DNS Lookup RTT Capture
• On-Path Telemetry methods (IOAM , AltMark )
• Estimation

Modeling full latency distributions may be too complex to allow for easy adoption of the framework. Instead, reporting latency at selected percentiles offers a practical compromise between accuracy and deployment considerations, trading off composability, which is only possible with distributions, for an easier deployment. A commonly accepted set of percentiles spanning from the 0th to the 100th in a logarithmic-like progression has been suggested by others and is recommended here: [0th, 10th, 25th, 50th, 75th, 90th, 95th, 99th, 99.9th, 100th]. The framework is agnostic to traffic direction but mandates that measurements specify whether latency is one-way or round-trip. To ensure broad applicability across diverse use cases, the QoO framework deliberately avoids prescribing specific conditions for sampling, such as fixed time intervals or defined network load levels. This flexibility enables deployment in both controlled and production environments. When measurements are taken during periods of network load, the result naturally includes latency under load. In scenarios such as passive monitoring of production traffic, capturing artificially loaded conditions may not always be feasible, whereas passively observing the actual network load may be possible. Importantly, the framework does not enforce a minimum sample count. This means that even a small number of samples (e.g., 10) could technically constitute a distribution but such cases are clearly insufficient for statistical confidence. The intent is to balance rigor with practicality, recognizing that constraints vary across devices, applications, and deployment environments. To support reproducibility and enable confidence analysis, each measurement must be accompanied by the following metadata:

• Description of the measurement path, including the endpoints (source and destination), network segments traversed, measurement points (if applicable), and direction (uplink, downlink, or bidirectional)
• Timestamp of first sample
• Total duration of the sampling period
• Number of samples collected
• Sampling method, including:
  • Cyclic: One sample every N milliseconds (specify N)
  • Burst: X samples every N milliseconds (specify X and N)
  • Passive: Opportunistic sampling of live traffic (non-uniform intervals)

These metadata elements are essential for interpreting the precision and reliability of the measurements. As demonstrated in , low sampling frequencies and short measurement durations can lead to misleadingly optimistic or imprecise QoO scores. To assess the precision of network performance measurements, implementers should consider:

• The repeatability of measurements under similar network conditions
• The impact of sampling frequency and duration on percentile estimates, particularly for high percentiles (e.g., 99th, 99.9th)
• The measurement uncertainty introduced by hardware/software timing jitter, clock synchronization errors, and other system-level noise sources
• The statistical confidence intervals for percentile estimates based on sample size

Acceptable levels of precision depend on the use case. Implementers should document their precision assessment methodology and report precision metrics alongside QoO scores when precision is critical for the use case.

Describing Network Performance Requirements The QoO framework expresses network performance requirements as a set of percentile-latency tuples with corresponding packet loss thresholds and a minimum required throughput. For example, a requirement might state: at 4Mbps, 90% of packets must arrive within 100ms, and 100% within 200ms, implying 0% packet loss. This list can be minimal (e.g., 100% within 200ms) or extended as needed and different percentiles may be used to characterize different applications. Still, it might be beneficial for future standardization activities to converge on a fixed set of general percentiles or for specific applications / application classes to make QoO measurements between different networks or providers more comparable. For the sake of simplicity, this document only states that the latency percentiles specified in the requirements must match the information provided by the measurements. This means that when the measurements report full distributions, requirements can use arbitrary percentiles while the requirement percentiles need to match one or more of the percentiles reported by the measurements in case the simplification described in is used, i.e., one can set requirements at the [0th, 10th, 25th, 50th, 75th, 90th, 95th, 99th, 99.9th, 100th] percentiles. Packet loss rates and bandwidth must be reported as separate values. Applications do of course have throughput requirements, and thus a complete framework for application-level network quality must also take capacity into account. Insufficient bandwidth may give unacceptable application outcomes without necessarily inducing a lot of latency or packet loss. Therefore, the network requirements must include a minimum throughput requirement. A fully specified requirement can be thought of as specifying the latency and loss requirements to be met while the end-to-end network path is loaded in a way that is at least as demanding of the network as the application itself. This may be achieved by running the actual application and measuring delay and loss alongside it, or by generating artificial traffic to a level at least equivalent to the application traffic load. The QoO framework focuses on latency, packet loss, and throughput under the assumption that other factors which could in principle be measured but are not explicitly captured by QoO will inevitably influence the observed latency and packet loss behavior, so that QoO indirectly accounts for their effects through the measured distributions. Whether the requirements are one-way or two-way must be specified. Where the requirement is one-way, the direction (user-to-network or network-to-user) must be specified. In case of a two-way requirement, a decomposition into uplink and downlink components may be specified. Network performance requirements and measurements are already standardized in the QED framework . This document extends the QED framework with a method that translates the network performance requirements and measurements into a network quality score that quantifies how close the provided network conditions are to the optimal conditions specified by the requirements. To that aim, first recall the key design goal of establishing a quantifiable distance between optimal and unacceptable network conditions, thereby enabling an objective assessment of relative quality. Accordingly, the requirements specification is extended to define both the network performance required for achieving optimal application performance and the lower network performance threshold below which the application performance is considered unacceptable. The two ends of the distance measure correspond to the Requirements for Optimal Performance (ROP) and the Conditions at the Point of Unacceptable Performance (CPUP). For example, ROP could be defined as: at 4Mbps, 99% of packets need to arrive within 100ms, 99.9% within 200ms, and 0.1% packet loss is acceptable for the outcome to be as intended. Similarly, CPUP could be defined as: if 99% of the packets have not arrived after 200ms, or 99.9% within 300ms, the perceived service will be unacceptable. If a latency percentile is included in the ROP, it must also be defined in the CPUP, and vice versa, i.e., neither specification should define a percentile that is not present in the other. For example, if the 99.9th percentile is part of the CPUP then the ROP must also include the 99.9th percentile. The derivation of ROP and CPUP values requires standardized testing conditions to ensure consistency and accuracy. Application developers should publish their testing methodologies, including the network conditions, hardware configurations, and measurement procedures used to establish these thresholds. Without such standardization, the overall accuracy and precision of QoO scores may be reduced due to variations in testing approaches across different applications and developers. Developers are encouraged to follow relevant standards for testing methodologies, such as ITU-T P-series recommendations for subjective quality assessment (, , ) and IETF IPPM standards for network performance measurement (, , ). These standards provide guidance on test design, measurement procedures, and statistical analysis that can help ensure consistent and reproducible threshold definitions. This document does not define a standardized approach for creating a quality-focused network performance requirement specification with only discussing general considerations for creating a specification.

Calculating QoO The QoO score compares the results of network performance measurements to application-specific network performance requirements by assessing how close the measured network performance is to the network conditions needed for optimal application performance, incorporating both latency and packet loss. There are three key scenarios:

• The network meets all requirements for optimal performance (ROP). QoO Score: 100%.
• The network fails one or more criteria for conditions at the point of unacceptable performance (CPUP). QoO Score: 0%.
• The network performance falls between optimal and unacceptable. In this case, a continuous QoO score between 0% and 100% is computed by taking the worst score derived from latency and packet loss.

Note that the QoO score should reflect the directionality of the measurements (one-way or round-trip) as specified in the network performance requirements. When comparing measurements to requirements, both must use the same directionality and, for one-way measurements, the same direction (uplink or downlink).

Overall QoO Calculation The overall QoO score is the minimum of a latency (QoO_latency) and a packet loss (QoO_loss) score: with QoO_latency and QoO_loss as defined in the following.

Latency Component The QoO latency score bases on linear interpolations of the latency values at all latency percentiles defined in ROP / CPUP and represents the minimum value for all percentiles: Where:

• latency_percentiles are the latency percentiles contained in the ROP and CPUP definitions
• ML[i] is the measured latency at percentile latency_percentiles[i].
• ROP[i] is the latency as indicated in the requirement for optimal performance at percentile latency_percentiles[i].
• CPUP[i] is the latency as indicated in the condition at the point of unacceptable performance at percentile latency_percentiles[i].

Packet Loss Component Packet loss is considered as a separate, single measurement that applies across the entire traffic sample, not at each percentile. The packet loss score is calculated using a similar interpolation formula, but based on the total measured packet loss (MLoss) and the packet loss thresholds defined in the ROP and CPUP: Where:

• M_Loss is the measured packet loss.
• ROP_Loss is the acceptable packet loss for optimal performance.
• CPUP_Loss is the packet loss threshold beyond which the application becomes unacceptable.

Example The following example illustrates the QoO calculations. Example requirements:

• Minimum bandwidth: 4 Mbps
• ROP: {99%, 200ms}, {99.9%, 300ms}, 1% packet loss
• CPUP: {99%, 500ms}, {99.9%, 600ms}, 5% packet loss

Example measured conditions:

• Measured latency: 99% = 350ms, 99.9% = 375ms
• Measured packet loss: 2%
• Measured minimum bandwidth: 32Mbps / 28Mbps

Latency component: Packet loss component: Overall QoO score: In this example, the network scores 50% on the QoO assessment range between unacceptable and optimal for the given application when using the measured network and considering both latency and packet loss. The score implies that the latency impact dominates the packet loss impact and that the network overall provides conditions at the midway point of the performance range.

Operational Considerations This document focuses on the core QoO framework and describes how quality-focused network performance requirements and corresponding network performance measurements can be used to calculate QoO network quality scores. Aiming to ensure broad and easy applicability of the QoO framework across diverse use cases, the document does not impose strict mandates. Instead, this section provides general guidance concerning the operation of the QoO framework based on intuitions and assumptions that guided the development of the framework. Future standardization activities are expected to capture and refine best practices once more operational experience has been gained.

Creating Network Performance Requirement Specifications This document does not define a standardized approach for creating a quality-focused network performance requirement specification. Instead, this section provides general guidance on and a rough outline for deriving an admittingly subjective requirement specification, aiming to create a basis for future standardization efforts focusing on developing a standardized, objective requirement creation framework. Additional information is provided in . Direct use of the approach described below in production scenarios is discouraged. When determining quality-focused network performance requirements for an application, the goal is to identify the network conditions where application performance is optimal and where it becomes unacceptable. There is no universally strict threshold at which network conditions render an application unacceptable. For optimal performance, some applications may have clear definitions, but for others, such as web browsing and gaming, lower latency and loss is always preferable. One approach for deriving possible thresholds is to run the application over a controlled network segment with adjustable quality and then vary the network conditions while continuously observing the resulting application-level performance. The latter can be assessed manually by the entity performing the testing or using automated methods, such as recording video stall duration within a video player. Additionally, application developers could set thresholds for acceptable fps, animation fluidity, i/o latency (voice, video, actions), or other metrics that directly affect the user experience and measure these user-facing metrics during tests to correlate the metrics with the network conditions. Using this scenario, one can first establish a baseline under excellent network conditions. Network conditions can then be gradually worsened by adding delay or packet loss or decreasing network capacity until the application no longer performs optimally. The corresponding network conditions identify the minimal requirements for optimal performance (ROP). Continuing to worsen the network conditions until the application fails completely eventually yields the network conditions at the point of unacceptable performance (CPUP). Note that different users may have different tolerance levels for application degradation. Hence, tests conducted by a single entity likely result in highly subjective thresholds. The thresholds established should represent acceptable performance for the target user base, which may require user studies or market research to determine appropriate values. As stated at the beginning of this section, this document does not define a standardized approach for creating a quality-focused network performance requirement specification and directly using the approach described above is discouraged.

Composability and Use Cases One of the key strengths of the QoO framework is the mathematical composability of the underlying Quality Attenuation metric, allowing network quality measurements to be combined across segments using convolution . This approach requires sharing the measured distributions for the involved segments among relevant stakeholders, which can be challenging across different operators or networks. However, even without sharing distributions across all networks of an end-to-end path, QoO remains valuable for analyzing and troubleshooting individual network segments. Operators can use QoO to assess specific segments within their own networks, and end-users can gain insights into their own connectivity as long as their network providers support QoO. Hence, QoO is well-suited for incremental deployment.

Deployment Considerations The QoO framework assumes that measurements reflect the actual connectivity service that will be provided to application flows. However, networks may offer multiple connectivity service levels (e.g., VPN services , corporate customer tiers, and network slicing configurations ). In such deployments, it is important to ensure that:

• Measurements are taken using the same connectivity service level that will be used by the application
• The measurement methodology accounts for any traffic prioritization, differentiated services, or quality-of-service mechanisms that may affect application performance
• Network configurations and policies that will apply to application traffic are reflected in the measurement conditions

Failing to align measurements with the actual service delivery may result in QoO scores that do not accurately reflect the application's expected performance.

Adaptive Applications Many modern applications are adaptive, meaning they can adjust their behavior based on network conditions. For example, video streaming applications may reduce bit rate when bandwidth is limited, or increase buffer size when latency is high. For adaptive applications, there are typically different levels of optimal performance rather than a single absolute threshold. For example, a video streaming application might provide different available video resolutions, ranging from 4K to 480p resolution. Combined with different transmission latencies, each of these resolutions can induce varying levels of perceived usability. The QoO framework can accommodate such applications by defining multiple ROP/CPUP thresholds corresponding to different quality levels. The framework can then assess how well the application will achieve each quality level, providing a more nuanced view of application performance than a simple binary pass/fail metric. Another, less complex approach at the cost of reduced fidelity in the QoO score, is to set the threshold for optimal performance at the highest rendition available for the video stream, and the threshold for unacceptability where the lowest rendition cannot be delivered without resulting in stalling events. Application developers implementing adaptive applications should consider publishing quality profiles that define network performance requirements for different adaptation levels, enabling more accurate QoO assessment.

Sensitity to Sampling Accuracy While the QoO framework itself places no strict requirement on sampling patterns or measurement technology, a simulation study conducted to inform the creation of this document examined the metric's real-world applicability under varying conditions and made the following conclusions:

1. Sampling Frequency: Slow sampling rates (e.g., <1Hz) risk missing rare, short-lived latency spikes, resulting in overly optimistic QoO scores.
2. Measurement Noise: Measurement errors on the same scale as the thresholds (ROP, CPUP) can distort high-percentile latencies and cause artificially lower QoO.
3. Requirement Specification: Slightly adjusting the latency thresholds or target percentiles can cause significant changes in QoO, especially when the measurement distribution is near a threshold.
4. Measurement Duration: Shorter tests with sparse sampling tend to underestimate worst-case behavior for heavy-tailed latency distributions, biasing QoO in a positive direction.

In summary, overly noisy or inaccurate latency samples can artificially inflate worst-case percentiles, thereby driving QoO scores lower than actual network conditions would warrant. Conversely, coarse measurement intervals can miss short-lived spikes entirely, resulting in an inflated QoO. From these findings, we deduce the following guidelines for practical application:

• Calibrate the combination of sampling rate and total measurement period to capture fat-tailed distributions of latency with sufficient accuracy.
• Avoid or account for significant measurement noise where possible (e.g., by calibrating time sources, accounting for clock drift, considering hardware/software measurement jitter).
• Thoroughly test application requirement thresholds so that the resulting QoO scores accurately reflect application performance.

These guidelines are non-normative but reflect empirical evidence on how QoO performs.

Insights From User Testing While subjective QoE testing as specified in the ITU-T P-series recommendations (, , and ) is out of scope of this document, a study involving 25 participants tested the QoO framework in real-world settings . Participants used specially equipped routers in their homes for ten days, providing both network performance data and feedback through pre- and post-trial surveys. Participants found QoO scores more intuitive and actionable than traditional metrics (e.g., speed tests). QoO directly aligned with their self-reported experiences, increasing trust and engagement. These results indicate that users find it easier to correlate QoO scores with real-world application performance than, for example, a speed test. As such, QoO is expected to help bridge technical metrics with application performance. However, the specific impact of QoO should be studied further, for example, via comparative studies with blinded methodologies that compare QoO to other QoS-type approaches or application-provided QoE ratings as the mentioned study's design might have introduced different forms of bias.

Known Weaknesses and Open Questions The described QoO framework simplifies the comparison between network performance requirements from applications and Quality Attenuation measurements. This simplification introduces several artifacts, the significance of which may vary depending on the context. The following section discusses some known limitations. A general assumption underlying the framework is that factors which could in principle be measured but are not explicitly captured by QoO (such as temporal packet ordering, fine-grained bandwidth variations, or the full shape of the latency distribution) will inevitably influence the observed latency and packet loss behavior, so that QoO indirectly accounts for their effects through the measured distributions.

Volatile Networks Volatile networks - in particular, mobile cellular networks - pose a challenge for network quality prediction, with the level of assurance of the prediction likely to decrease as session duration increases. Historic network conditions for a given cell may help indicate times of network load or reduced transmission power, and their effect on throughput/latency/loss. However, as terminals are mobile, the signal bandwidth available to a given terminal can change by an order of magnitude within seconds due to physical radio factors. These include whether the terminal is at the edge of a cell for a radio network, or undergoing cell handover, the radio interference and fading from the local environment, and any switch between radio bearers with differing signal bandwidth and transmission-time intervals (e.g., 3GPP 4G and 5G). This suggests a requirement for measuring Quality Attenuation to and from an individual terminal, as that can account for the factors described above. How that facility is provisioned onto individual terminals and how terminal-hosted applications can trigger a Quality Attenuation query, is an open question.

Missing Temporal Information in Distributions. The two latency series (1,200,1,200,1,200,1,200,1,200) and (1,1,1,1,1,200,200,200,200,200) have identical distributions, but may have different application performance. Ignoring this information is a tradeoff between simplicity and precision. To capture all information necessary to adequately capture outcomes quickly gets into extreme levels of overhead and high computational complexity. An application's performance depends on reactions to varying network conditions, meaning nearly all different series of latencies may have different application outcomes.

Subsampling the Real Distribution Additionally, it is not feasible to capture latency for every packet transmitted. Probing and sampling can be performed, but some aspects will always remain unknown. This introduces an element of uncertainty and perfect predictions cannot be achieved; rather than disregarding this reality, it is more practical to acknowledge it. Therefore, discussing the assessment of outcomes provides a more accurate and meaningful approach.

Assuming Linear Relationship Between Optimal Performance and Unusable It has been shown that, for example, interactivity cannot be modeled by a linear scale . Thus, the linear modeling proposed here adds an error in estimating the perceived performance of interactive applications. One can conjure up scenarios where 50ms latency is actually worse than 51ms latency as developers may have chosen 50ms as the threshold for changing quality, and the threshold may be imperfect. Taking these scenarios into account would add another magnitude of complexity to determining network performance requirements and finding a distance measure (between requirement and actual measured capability).

Binary Bandwidth Threshold Choosing a binary bandwidth threshold is to reduce complexity, but it must be acknowledged that many applications are not that simple. Network requirements can be set up per quality level (resolution, frames per-second, etc.) for the application if necessary.

Arbitrary Selection of Percentiles A selection of percentiles is necessary for simplicity, because more complex methods may slow adoption of the framework. The 0th (minimum) and 50th (median) percentiles are commonly used for their inherent significance. According to , the 90th, 98th, and 99th percentiles are particularly important for certain applications. Generally, higher percentiles provide more insight for interactive applications, but only up to a certain threshold beyond which applications may treat excessive delays as packet loss and adapt accordingly. The choice between percentiles such as the 95th, 96th, 96.5th, or 97th is not universally prescribed and may vary between application types. Therefore, percentiles must be selected arbitrarily, based on the best available knowledge and the intended use case.

Security Considerations The QoO framework introduces a method for assessing network quality based on probabilistic outcomes derived from latency, packet loss, and throughput measurements. While the framework itself is primarily analytical and does not define a new protocol, some security considerations arise from its deployment and use.

Measurement Integrity and Authenticity QoO relies upon accurate and trustworthy measurements of network performance. If an attacker can manipulate these measurements, either by injecting falsified data or tampering with the measurement process, they could distort the resulting QoO scores. This could mislead users, operators, or regulators into making incorrect assessments of network quality. To mitigate this risk:

• Measurement agents have to authenticate with the systems collecting or analyzing QoO data.
• Measurement data has to be transmitted over secure channels (e.g., (D)TLS) to ensure confidentiality and integrity.
• Digital signatures may be used to verify the authenticity of measurement reports.

Risk of Misuse and Gaming As QoO scores may influence regulatory decisions, SLAs, or user trust, there is a risk that network operators or application developers might attempt to "game" the system. For example, they might optimize performance only for known test conditions or falsify requirement thresholds to inflate QoO scores. Mitigations include:

• Independent verification of application requirements and measurement methodologies.
• Use of randomized testing procedures.
• Transparency in how QoO scores are derived and what assumptions are made.

Denial-of-Service (DoS) Risks Active measurement techniques used to gather QoO data (e.g., TWAMP, STAMP, and synthetic traffic generation) can place additional load on a network. If not properly rate-limited, this may inadvertently degrade services offered by a network or be exploited by malicious actors to launch DoS attacks. To mitigate these risks, the following is recommended:

• Implement rate-limiting and access control for active measurement tools.
• Ensure that measurement traffic does not interfere with critical services.
• Monitor for abnormal measurement patterns that may indicate abuse.

Trust in Application Requirements QoO depends on application developers to define ROP and CPUP. If these are defined inaccurately-either unintentionally or maliciously-the resulting QoO scores may be misleading. To address such risks, the following recommendations are made:

• Encourage peer review and publication of application requirement profiles.
• Where QoO is used for regulatory or SLA enforcement, require independent validation of requirement definitions.

Privacy Considerations QoO measurements may involve collecting detailed performance data from end-user devices or applications. Depending on the deployment model, this includes metadata such as IP addresses, timestamps, or application usage patterns. To protect user privacy:

• Data collection should be subject to user consent prior to collecting data.
• Data collection should follow the principle of data minimization, only collecting what is strictly necessary.
• Privacy-sensitive information (e.g., Personally Identifiable Information (PII)) should be anonymized or pseudonymized where possible.
• Users should be informed about what data is collected and how it is used, in accordance with applicable privacy regulations (e.g., General Data Protection Regulation (GDPR)).

Comparison To Other Network Quality Metrics Numerous network quality metrics and associated frameworks have been proposed, adopted, and, at times, misapplied over the years. The following is a brief overview of several key network quality metrics in comparison to QoO. Each metric is evaluated against the three criteria established in . summarizes the properties of each of the surveyed metrics. Summary of Performance Metrics Properties

Metric	Can Assess How Well Applications Are Expected to Work	Easy to Articulate Application Requirements	Composable
Throughput	Yes for some applications	Yes	No
Mean latency	Yes for some applications	Yes	Yes
99th Percentile of Latency	No	No	No
Variance of latency	No	No	Yes
IPDV	Yes for some applications	No	No
PDV	Yes for some applications	No	No
Trimmed mean of latency	Yes for some applications	Yes	No
Round Trips Per Minute	Yes for some applications	Yes	No
Quality Attenuation	Yes	No	Yes
Quality of Outcome	Yes	Yes	Yes (Through Quality Attenuation)

The column "Can Assess How Well Applications Are Expected to Work" indicates whether a metric can, in principle, capture relevant information to assess application performance, assuming that measurements cover the properties of the end-to-end network path that the application uses. "Easy to Articulate Application Requirements" refers to the ease with which application-specific requirements can be expressed using the respective metric. "Composable" indicates whether the metric supports mathematical composition to enable detailed network analysis.

Throughput Throughput is related to user-observable application outcomes because there must be enough bandwidth available. Adding extra bandwidth above a certain threshold will, at best, receive diminishing returns (and any returns are often due to reduced latency). It is not possible to assess optimal or unacceptable application performance based on throughput alone for most applications. Throughput can be compared to a variety of application requirements, but since there is no direct correlation between throughput and application performance, it is not possible to conclude that an application will work well even if it is known that enough throughput is available. Throughput cannot be composed.

Mean Latency Mean latency relates to user-observable application outcomes in the sense that the mean latency must be low enough to support a good experience. However, it is not possible to conclude that a general application will work well if the mean latency is good enough . Mean latency can be composed. For example, if the mean latency values of links a-b and b-c are known, then the mean latency of the composition a-b-c is the sum of a-b and b-c.

99th Percentile of Latency The 99th percentile of latency relates to user-observable application outcomes because it captures some information about how bad the tail latency is. If an application can handle 1% of packets being too late, for instance by maintaining a playback buffer, then the 99th percentile can be a good metric for measuring application performance. It does not work as well for applications that are very sensitive to overly delayed packets because the 99th percentile disregards all information about the delays of the worst 1% of packets. It is not possible to compose 99th-percentile values.

Variance of Latency The variance of latency can be calculated from any collection of samples, but network latency is not necessarily normally distributed. As such, it can be difficult to extrapolate from a measure of the variance of latency to how well specific applications will work. The variance of latency can be composed. For example, if the variance values of links a-b and b-c are known, then the variance of the composition a-b-c is the sum of the variances a-b and b-c.

Inter-Packet Delay Variation (IPDV) The most common definition of IPDV measures the difference in one-way delay between subsequent packets. Some applications are very sensitive to this performance characteristic because of time-outs that cause later-than-usual packets to be discarded. For some applications, IPDV can be useful in assessing application performance, especially when it is combined with other latency metrics. IPDV does not contain enough information to assess how well a wide range of applications will work. IPDV cannot be composed.

Packet Delay Variation (PDV) The most common definition of PDV measures the difference in one-way delay between the smallest recorded latency and each value in a sample. PDV cannot be composed.

Trimmed Mean of Latency The trimmed mean of latency is the mean computed after the worst x percent of samples have been removed. Trimmed means are typically used in cases where there is a known rate of measurement errors that should be filtered out before computing results. In the case where the trimmed mean simply removes measurement errors, the result can be composed in the same way as the mean latency. In cases where the trimmed mean removes real measurements, the trimming operation introduces errors that may compound when composed.

Round-trips Per Minute Round-trips per minute is a metric and test procedure specifically designed to measure delays as experienced by application-layer protocol procedures, such as HTTP GET, establishing a TLS connection, and DNS lookups. Hence, it measures something very close to the user-perceived application performance of HTTP-based applications. RPM loads the network before conducting latency measurements and is, therefore, a measure of loaded latency (also known as working latency), and well-suited to detecting bufferbloat . RPM is not composable.

Quality Attenuation Quality Attenuation is a network quality metric that combines dedicated latency and packet loss distributions into a single variable . It relates to user-observable outcomes in the sense that they can be measured using the Quality Attenuation metric directly, or the Quality Attenuation value describing the time-to-completion of a user-observable outcome can be computed if the Quality Attenuation of each sub-goal required to reach the desired outcome is known . Quality Attenuation is composable because the convolution of Quality Attenuation values allows computing the time it takes to reach specific outcomes given the Quality Attenuation of each sub-goal and the causal dependency conditions between them .

Quality of Outcome Quality of Outcome (QoO) builds upon Quality Attenuation by adding application-specific, dual-threshold network performance requirements (ROP and CPUP) and translating the comparison between measured network conditions and these requirements into a percentage-based score. By incorporating latency distributions, packet loss rates, and throughput measurements, QoO can assess how well a wide range of applications are expected to perform under given network conditions. The underlying Quality Attenuation measurements used in QoO are mathematically composable, as latency distributions can be composed using convolution . This composability extends to QoO in the sense that operators can measure individual network segments, compose the underlying Quality Attenuation distributions, and then compute QoO scores from the composed result.

IANA Considerations This document has no IANA actions.

Implementation status Note to RFC Editor: This section must be removed before publication of the document. This section records the status of known implementations of the protocol defined by this specification at the time of posting of this Internet-Draft, and is based on a proposal described in . The description of implementations in this section is intended to assist the IETF in its decision processes in progressing drafts to RFCs. Please note that the listing of any individual implementation here does not imply endorsement by the IETF. Furthermore, no effort has been spent to verify the information presented here that was supplied by IETF contributors. This is not intended as, and must not be construed to be, a catalog of available implementations or their features. Readers are advised to note that other implementations may exist. According to , "this will allow reviewers and working groups to assign due consideration to documents that have the benefit of running code, which may serve as evidence of valuable experimentation and feedback that have made the implemented protocols more mature. It is up to the individual working groups to use this information as they see fit".

qoo-c

• Link to the open-source repository: https://github.com/getCUJO/qoo-c
• The organization responsible for the implementation: CUJO AI
• A brief general description: A C library for calculating Quality of Outcome
• The implementation's level of maturity: A complete implementation of the specification described in this document
• Coverage: The library is tested with unit tests
• Licensing: MIT
• Implementation experience: Tested by the author. Needs additional testing by third parties.
• Contact information: Bjørn Ivar Teigen Monclair: bjorn.monclair@cujo.com
• The date when information about this particular implementation was last updated: 27th of May 2025

goresponsiveness

• Link to the open-source repository: https://github.com/network-quality/goresponsiveness The specific pull-request: https://github.com/network-quality/goresponsiveness/pull/56
• The organization responsible for the implementation: University of Cincinatti for goresponsiveness as a whole, Domos for the QoO part.
• A brief general description: A network quality test written in Go. Capable of measuring RPM and QoO.
• The implementation's level of maturity: Under active development; partial QoO support integrated.
• Coverage: The QoO part is tested with unit tests
• Licensing: GPL 2.0
• Implementation experience: Needs testing by third parties
• Contact information: Bjørn Ivar Teigen Monclair: bjorn.monclair@cujo.com William Hawkins III: hawkinwh@ucmail.uc.edu
• The date when information about this particular implementation was last updated: 10th of January 2024

References Normative References This memo utilizes IP performance metrics that are applicable to both complete paths and sub-paths, and it defines relationships to compose a complete path metric from the sub-path metrics with some accuracy with regard to the actual metrics. This is called "spatial composition" in RFC 2330. The memo refers to the framework for metric composition, and provides background and motivation for combining metrics to derive others. The descriptions of several composed metrics and statistics follow. [STANDARDS-TRACK] This document describes a framework and a process for developing Performance Metrics of protocols and applications transported over IETF-specified protocols. These metrics can be used to characterize traffic on live networks and services. This memo documents an Internet Best Current Practice. Broadband Forum Informative References ISO BITAG This memo defines a metric for round-trip delay of packets across Internet paths. [STANDARDS-TRACK] The One-way Active Measurement Protocol (OWAMP), specified in RFC 4656, provides a common protocol for measuring one-way metrics between network devices. OWAMP can be used bi-directionally to measure one-way metrics in both directions between two network elements. However, it does not accommodate round-trip or two-way measurements. This memo specifies a Two-Way Active Measurement Protocol (TWAMP), based on the OWAMP, that adds two-way or round-trip measurement capabilities. The TWAMP measurement architecture is usually comprised of two hosts with specific roles, and this allows for some protocol simplifications, making it an attractive alternative in some circumstances. [STANDARDS-TRACK] Packet delay variation metrics appear in many different standards documents. The metric definition in RFC 3393 has considerable flexibility, and it allows multiple formulations of delay variation through the specification of different packet selection functions. Although flexibility provides wide coverage and room for new ideas, it can make comparisons of independent implementations more difficult. Two different formulations of delay variation have come into wide use in the context of active measurements. This memo examines a range of circumstances for active measurements of delay variation and their uses, and recommends which of the two forms is best matched to particular conditions and tasks. This memo provides information for the Internet community. Many user applications (and the transport protocols that make them possible) require two-way communications. To assess this capability, and to achieve test system simplicity, round-trip loss measurements are frequently conducted in practice. The Two-Way Active Measurement Protocol specified in RFC 5357 establishes a round-trip loss measurement capability for the Internet. However, there is currently no round-trip packet loss metric specified according to the RFC 2330 framework. This memo adds round-trip loss to the set of IP Performance Metrics (IPPM). [STANDARDS-TRACK] This memo defines a metric for one-way delay of packets across Internet paths. It builds on notions introduced and discussed in the IP Performance Metrics (IPPM) Framework document, RFC 2330; the reader is assumed to be familiar with that document. This memo makes RFC 2679 obsolete. This memo defines a metric for one-way loss of packets across Internet paths. It builds on notions introduced and discussed in the IP Performance Metrics (IPPM) Framework document, RFC 2330; the reader is assumed to be familiar with that document. This memo makes RFC 2680 obsolete. This memo provides clear definitions for Active and Passive performance assessment. The construction of Metrics and Methods can be described as either "Active" or "Passive". Some methods may use a subset of both Active and Passive attributes, and we refer to these as "Hybrid Methods". This memo also describes multiple dimensions to help evaluate new methods as they emerge. This document describes a simple process that allows authors of Internet-Drafts to record the status of known implementations by including an Implementation Status section. This will allow reviewers and working groups to assign due consideration to documents that have the benefit of running code, which may serve as evidence of valuable experimentation and feedback that have made the implemented protocols more mature. This process is not mandatory. Authors of Internet-Drafts are encouraged to consider using the process for their documents, and working groups are invited to think about applying the process to all of their protocol specifications. This document obsoletes RFC 6982, advancing it to a Best Current Practice. Bufferbloat is a phenomenon in which excess buffers in the network cause high latency and latency variation. As more and more interactive applications (e.g., voice over IP, real-time video streaming, and financial transactions) run in the Internet, high latency and latency variation degrade application performance. There is a pressing need to design intelligent queue management schemes that can control latency and latency variation, and hence provide desirable quality of service to users. This document presents a lightweight active queue management design called "PIE" (Proportional Integral controller Enhanced) that can effectively control the average queuing latency to a target value. Simulation results, theoretical analysis, and Linux testbed results have shown that PIE can ensure low latency and achieve high link utilization under various congestion situations. The design does not require per-packet timestamps, so it incurs very little overhead and is simple enough to implement in both hardware and software. The purpose of this informational document is to establish test and evaluation methodology and measurement techniques for physical network equipment in the data center. RFC 8238 is a prerequisite for this document, as it contains terminology that is considered normative. Many of these terms and methods may be applicable beyond the scope of this document as the technologies originally applied in the data center are deployed elsewhere. This memo presents the FQ-CoDel hybrid packet scheduler and Active Queue Management (AQM) algorithm, a powerful tool for fighting bufferbloat and reducing latency. FQ-CoDel mixes packets from multiple flows and reduces the impact of head-of-line blocking from bursty traffic. It provides isolation for low-rate traffic such as DNS, web, and videoconferencing traffic. It improves utilisation across the networking fabric, especially for bidirectional traffic, by keeping queue lengths short, and it can be implemented in a memory- and CPU-efficient fashion across a wide range of hardware. This document summarizes an operator's perception of the benefits that may be provided by intermediary devices that execute functions beyond normal IP forwarding. Such intermediary devices are often called "middleboxes". RFC 3234 defines a taxonomy of middleboxes and issues in the Internet. Most of those middleboxes utilize or modify application- layer data. This document primarily focuses on devices that observe and act on information carried in the transport layer, and especially information carried in TCP packets. This document describes the Simple Two-way Active Measurement Protocol (STAMP), which enables the measurement of both one-way and round-trip performance metrics, like delay, delay variation, and packet loss. This document defines the core of the QUIC transport protocol. QUIC provides applications with flow-controlled streams for structured communication, low-latency connection establishment, and network path migration. QUIC includes security measures that ensure confidentiality, integrity, and availability in a range of deployment circumstances. Accompanying documents describe the integration of TLS for key negotiation, loss detection, and an exemplary congestion control algorithm. In situ Operations, Administration, and Maintenance (IOAM) collects operational and telemetry information in the packet while the packet traverses a path between two points in the network. This document discusses the data fields and associated data types for IOAM. IOAM-Data-Fields can be encapsulated into a variety of protocols, such as Network Service Header (NSH), Segment Routing, Generic Network Virtualization Encapsulation (Geneve), or IPv6. IOAM can be used to complement OAM mechanisms based on, e.g., ICMP or other types of probe packets. This document discusses manageability of the QUIC transport protocol and focuses on the implications of QUIC's design and wire image on network operations involving QUIC traffic. It is intended as a "user's manual" for the wire image to provide guidance for network operators and equipment vendors who rely on the use of transport-aware network functions. The Measuring Network Quality for End-Users workshop was held virtually by the Internet Architecture Board (IAB) on September 14-16, 2021. This report summarizes the workshop, the topics discussed, and some preliminary conclusions drawn at the end of the workshop. Note that this document is a report on the proceedings of the workshop. The views and positions documented in this report are those of the workshop participants and do not necessarily reflect IAB views and positions. This document describes the Alternate-Marking technique to perform packet loss, delay, and jitter measurements on live traffic. This technology can be applied in various situations and for different protocols. According to the classification defined in RFC 7799, it could be considered Passive or Hybrid depending on the application. This document obsoletes RFC 8321. Computing in the Network (COIN) comes with the prospect of deploying processing functionality on networking devices such as switches and network interface cards. While such functionality can be beneficial, it has to be carefully placed into the context of the general Internet communication, and it needs to be clearly identified where and how those benefits apply. This document presents some use cases to demonstrate how a number of salient COIN-related applications can benefit from COIN. Furthermore, to guide research on COIN, it identifies essential research questions and outlines desirable capabilities that COIN systems addressing these use cases may need to support. Finally, the document provides a preliminary categorization of the described research questions to source future work in this domain. This document is a product of the Computing in the Network Research Group (COINRG). It is not an IETF product and it is not a standard. Everything OPS Cisco Old Dog Consulting Nokia Blue Fern Consulting ZTE New Protocols or Protocol Extensions are best designed with due consideration of the functionality needed to operate and manage them. Retrofitting operations and management considerations is suboptimal. The purpose of this document is to provide guidance to authors and reviewers on what operational and management aspects should be addressed when defining New Protocols or Protocol Extensions. This document obsoletes RFC 5706, replacing it completely and updating it with new operational and management techniques and mechanisms. It also updates RFC 2360 to obsolete mandatory MIB creation and introduces a requirement to include an "Operational Considerations" section in new RFCs in the IETF Stream. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. Microsoft n.d. Worcester Polytechnic Institute,Worcester,MA,USA Worcester Polytechnic Institute,Worcester,MA,USA Worcester Polytechnic Institute,Worcester,MA,USA IEEE ITU-T ITU-T ITU-T ITU-T ITU-T ITU-T This framework describes a practical methodology for measuring end- to-end TCP Throughput in a managed IP network. The goal is to provide a better indication in regard to user experience. In this framework, TCP and IP parameters are specified to optimize TCP Throughput. This document is not an Internet Standards Track specification; it is published for informational purposes. This document describes a framework for Virtual Private Networks (VPNs) running across IP backbones. This memo provides information for the Internet community. This document describes network slicing in the context of networks built from IETF technologies. It defines the term "IETF Network Slice" to describe this type of network slice and establishes the general principles of network slicing in the IETF context. The document discusses the general framework for requesting and operating IETF Network Slices, the characteristics of an IETF Network Slice, the necessary system components and interfaces, and the mapping of abstract requests to more specific technologies. The document also discusses related considerations with monitoring and security. This document also provides definitions of related terms to enable consistent usage in other IETF documents that describe or use aspects of IETF Network Slices.

Acknowledgments The authors would like to thank Karl Magnus Kalvik, Olav Nedrelid, and Knut Joar Strømmen for their conceptual and technical contributions to the QoO framework. The authors would like to thank Gavin Young, Brendan Black, Kevin Smith, Gino Dion, Mayur Sarode, Greg Mirsky, Wim De Ketelaere, Peter Thompson, and Neil Davies for their contributions to the initial version of this document. The authors would like to thank Paul Aitken, Mohamed Boucadair, Stuart Cheshire, Neil Davies, Guiseppe Fioccola, Ruediger Geib, Will Hawkins, Marcus Ihlar, Mehmet Şükrü Kuran, Paul Kyzivat, Jason Livingood, Greg Mirsky, Tal Mizrahi, Luis Miguel Contreras Murillo, Tommy Pauly, Alexander Raake, Werner Robitza, Kevin Smith, Martin Thomson, and Michael Welzl for their feedback and input to this document throughout the IETF process.