]> China Mobile

yaokehan@chinamobile.com

Telefonica

luismiguel.contrerasmurillo@telefonica.com

Huawei Technologies

shihang9@huawei.com

China Unicom

zhangs366@chinaunicom.cn

Alibaba Group

anqing.aq@alibaba-inc.com

cats Distributed computing is a computing pattern that service providers can follow and use to achieve better service response time and optimized energy consumption. In such a distributed computing environment, compute intensive and delay sensitive services can be improved by utilizing computing resources hosted in various computing facilities. Ideally, compute services are balanced across servers and network resources to enable higher throughput and lower response time. To achieve this, the choice of server and network resources should consider metrics that are oriented towards compute capabilities and resources instead of simply dispatching the service requests in a static way or optimizing solely on connectivity metrics. The process of selecting servers or service instance locations, and of directing traffic to them on chosen network resources is called "Computing-Aware Traffic Steering" (CATS). This document provides the problem statement and the typical scenarios for CATS, which shows the necessity of considering more factors when steering traffic to the appropriate computing resource to better meet the customer's expectations.

Introduction Over-the-top services depend on Content Delivery Networks (CDNs) for service provisioning, which has become a driving force for network operators to extend their capabilities by complementing their network infrastructure with CDN capabilities, particularly in edge sites. In addition, more computing resources are deployed at these edge sites as well. The fast development of this converged network and compute infrastructure is driven by user demands. On one hand, users want the best experience, e.g., expressed in low latency and high reliability, for new emerging applications such as high-definition video, Augmented Reality(AR)/Virtual Reality(VR), live broadcast. On the other hand, users want stable experience when moving among different areas. Generally, edge computing aims to provide better response time and transfer rates compared to cloud computing, by moving the computing towards the edge of a network. There are millions of home gateways, thousands of base stations, and hundreds of central offices in a city that could serve as compute-capable nodes to deliver a service. Note that not all of these nodes would be considered as edge nodes in some views of the network, but they can all provide computing resources for services. It brings some key problems on service deployment and traffic scheduling to the most suitable computing resource in order to meet users' demands. A service instance deployed at a single site might not provide sufficient capacity to fully guarantee the quality of service required by a customer. Especially at peak hours, computing resources at a single site can not handle all the incoming service requests, leading to longer response time or even dropping of requests experienced by clients. Moreover, increasing the computing resources hosted at each location to the potential maximum capacity is neither feasible nor economically viable in many cases. Offloading compute intensive processing to the user devices is not acceptable, since it would place pressure on local resources such as the battery and incur some data privacy issues if the needed data for computation is not provided locally. Instead, the same service can be deployed at multiple sites for better availability and scalability. Furthermore, it is desirable to balance the load across all service instances to improve throughput. For this, traffic needs to be steered to the 'best' service instance according to information that may include current computing load, where the notion of 'best' may highly depend on the application demands. This document describes sample usage scenarios that drive CATS requirements and will help to identify candidate solution architectures and solutions.

Definition of Terms This document uses the terms defined in .

Service identifier:

An identifier representing a service, which the clients use to access it.

Network Edge:

The network edge is an architectural demarcation point used to identify physical locations where the corporate network connects to third-party networks.

Edge Computing:

Edge computing is a computing pattern that moves computing infrastructures, i.e, servers, away from centralized data centers and instead places it close to the end users for low latency communication. Relations with network edge: edge computing infrastructures connect to corporate network through a network edge entry/exit point.

Even though this document is not a protocol specification, it makes use of upper case key words to define requirements unambiguously. The key words "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.

Problem Statement

Multi-deployment of Edge Sites and Service Since edge computing aims at a closer computing service based on the shorter network path, there will be more than one edge site with the same application in the city/province/state, a number of representative cities have deployed multi-edge sites and the typical applications, and there are more edge sites to be deployed in the future. Before deploying edge sites, there are some factors that need to be considered, such as:

• The existing infrastructure capacities, which could be updated to edge sites, e.g. operators' machine room.
• The amount and frequency of computing resource that is needed.
• The network resource status linked to computing resource.

To improve the effectiveness of service deployment, the problem of how to choose the optimal edge node on which to deploy services needs to be solved. introduces considerations for how to deploy applications or functions to the edge, such as the type of instance, optional storage extension, optional hardware acceleration characteristics, and the compute flavor of CPU/GPU, etc. More network and service factors may also be considered, such as:

• Network and computing resource topology: The overall consideration of network access, connectivity, path protection or redundancy, and the location and overall distribution of computing resources in the network, and the relative position within the network topology.
• Location: The number of users, the differentiation of service types, and the number of connections requested by users, etc. For edge nodes located in populous area with a large number of users and service requests, service duplication could be deployed more than in other areas.
• Capacity of multiple edge nodes: Not only the capacity of a single node, but also the total number of requests that can be processed by the resource pool composed of multiple nodes.
• Service category: For example, whether the service is a multi-user interaction, such as video conferencing, or games, or whether it just resource acquisition, such as video viewing. ALTO can help to obtain one or more of the above pieces of information, so as to provide suggestions or formulate principles and strategies for service deployment.

This information could be collected periodically, and could record the total consumption of computing resources, or the total number of sessions accessed. This would indicate whether additional service instances need to be deployed. Unlike the scheduling of service requests, service deployment should follow the principle of proximity to place new service instances near to customer sites that will request them. If the resources are insufficient to support new instances, the operator can be informed to increase the hardware resources. In general, the choice of where to locate service instances and when to create new ones in order to provide the right levels of resource to support user demands is important in building a network that supports computing services. However, those aspects are out of scope for CATS and are left for consideration in another document.

Traffic Steering among Edges Sites and Service Instances This section describes how existing edge computing systems do not provide all of the support needed for real-time or near-real-time services, and how it is necessary to steer traffic to different sites considering mobility of people, different time slots, events, server loads, network capabilities, and some other factors which might not be directly measured, i.e., properties of edge sites(e.g., geographical location), etc. In edge computing, the computing resources and network resources are considered when deploying edge sites and services. Traffic is steered to an edge site that is "closest" or to one of a few "close" sites using load-balancing. But the "closest" site is not always the "best" as the status of computing resources and of the network may vary as follows:

• Closest site may not have enough resource, the load may dynamically change.
• Closest site may not have related resource, heterogeneous hardware in different sites.
• The network path to the closest site might not provide the necessary network characteristics, such as low latency or high throughput.

To address these issues some enhancements are needed to steer traffic to sites that can support the requested services. We assume that clients access one or more services with an objective to meet a desired user experience. Each participating service may be realized at one or more places in the network (called, service instances). Such service instances are instantiated and deployed as part of the overall service deployment process, e.g., using existing orchestration frameworks, within so-called edge sites, which in turn are reachable through a network infrastructure via an edge router. When a client issues a service request for a required service, the request is steered to one of the available service instances. Each service instance may act as a client towards another service, thereby seeing its own outbound traffic steered to a suitable service instance of the request service and so on, achieving service composition and chaining as a result. The aforementioned selection of one of candidate service instances is done using traffic steering methods, where the steering decision may take into account pre-planned policies (assignment of certain clients to certain service instances), realize shortest-path to the 'closest' service instance, or utilize more complex and possibly dynamic metric information, such as load of service instances, latency experienced or similar, for a more dynamic selection of a suitable service instance. It is important to note that clients may move. This means that the service instance that was "best" at one moment might no longer be best when a new service request is issued. This creates a (physical) dynamicity that will need to be catered for in addition to the changes in server and network load. shows a common way to deploy edge sites in the metro. There is an edge data center for metro area which has high computing resource and provides the service to more User Equipments(UEs) at the working time. Because more office buildings are in the metro area. And there are also some remote edge sites which have limited computing resource and provide the service to the UEs closed to them. Applications to meet service demands could be deployed in both the edge data center in metro area and the remote edge sites. In this case, the service request and the resource are matched well. Some potential traffic steering may be needed just for special service request or some small scheduling demand.

Common Deployment of Edge Sites

shows that during non-working hours, for example at weekend or daily night, more UEs move to the remote area that are close to their house or for some weekend events. So there will be more service request at remote but with limited computing resource, while the rich computing resource might not be used with less UE in the metro area. It is possible for many people to request services at the remote area, but with the limited computing resource, moreover, as the people move from the metro area to the remote area, the edge sites that serve common services will also change, so it may be necessary to steer some traffic back to the metro data center.

Steering Traffic among Edge Sites

There will also be the common variable of network and computing resources, for someone who is not moving but get a poor latency sometime. Because of other UEs moving, a large number of request for temporary events such as vocal concert, shopping festival and so on, and there will also be the normal change of the network and computing resource status. So for some fixed UEs, it is also expected to steer the traffic to appropriate sites dynamicity. Those problems indicate that traffic needs to be steered among different edge sites, because of the mobility of the UE and the common variable of network and computing resources. Moreover, some use cases in the following section require both low latency and high computing resource usage or specific computing hardware capabilities (such as local GPU); hence joint optimization of network and computing resource is needed to guarantee the Quality of Experience (QoE).

Use Cases This section presents a non-exhaustive set of use cases which would benefit from the dynamic selection of service instances and the steering of traffic to those service instances.

Example 1: Computing-aware AR or VR Cloud VR/AR introduces the concept of cloud computing to the rendering of audiovisual assets in such applications. Here, the edge cloud helps encode/decode and render content. The end device usually only uploads posture or control information to the edge and then VR/AR contents are rendered in the edge cloud. The video and audio outputs generated from the edge cloud are encoded, compressed, and transmitted back to the end device or further transmitted to central data center via high bandwidth networks. A Cloud VR service is delay-sensitive and influenced by both network and computing resources. Therefore, the edge node which executes the service has to be carefully selected to make sure it has sufficient computing resource and good network condition to guarantee the end-to-end service delay. For example, for an entry-level cloud VR (panoramic 8K 2D video) with 110-degree Field of View (FOV) transmission, the typical network requirements are bandwidth 40Mbps, 20ms for motion-to-photon latency, packet loss rate is 2.4E-5; the typical computing requirements are 8K H.265 real-time decoding, 2K H.264 real-time encoding. We can further divide the 20ms latency budget into:

1. sensor sampling delay(client), which is considered imperceptible by users is less than 1.5ms including an extra 0.5ms for digitalization and end device processing.
2. display refresh delay(client), which take 7.9ms based on the 144Hz display refreshing rate and 1ms extra delay to light up.
3. image/frame rendering delay(server), which could be reduced to 5.5ms.
4. round trip network delay(network), which should be bounded to 20-1.5-5.5-7.9 = 5.1ms.

So the budgets for server(computing) delay and network delay are almost equivalent, which make sense to consider both of the delay for computing and network. And it can't meet the total delay requirements or find the best choice by either optimize the network or computing resource. Based on the analysis, here are some further assumption as shows, the client could request any service instance among 3 edge sites. The delay of client could be same, and the differences of different edge sites and corresponding network path has different delays:

• Edge site 1: The computing delay=4ms based on a light load, and the corresponding network delay=9ms based on a heavy traffic.
• Edge site 2: The computing delay=10ms based on a heavy load, and the corresponding network delay=4ms based on a light traffic.
• Edge site 3: The edge site 3's computing delay=5ms based on a normal load, and the corresponding network delay=5ms based on a normal traffic.

In this case, we can't get a optimal network and computing total delay if choose the resource only based on either of computing or network status:

• If choosing the edge site based on the best computing delay it will be the edge site 1, the E2E delay=22.4ms.
• If choosing the edge site based on the best network delay it will be the edge site 2, the E2E delay=23.4ms.
• If choosing the edge site based on both of the status it will be the edge site 3, the E2E delay=19.4ms.

So, the best choice to ensure the E2E delay is edge site 3, which is 19.4ms and is less than 20ms. The differences of the E2E delay is only 3~4ms among the three, but some of them will meet the application demand while some doesn't. The conclusion is that it requires to dynamically steer traffic to the appropriate edge to meet the E2E delay requirements considering both network and computing resource status. Moreover, the computing resources have a big difference in different edges, and the "closest site" may be good for latency but lacks GPU support and should therefore not be chosen.

Computing-Aware AR or VR

Furthermore, specific techniques may be employed to divide the overall rendering into base assets that are common across a number of clients participating in the service, while the client-specific input data is being utilized to render additional assets. When being delivered to the client, those two assets are being combined into the overall content being consumed by the client. The requirements for sending the client input data as well as the requests for the base assets may be different in terms of which service instances may serve the request, where base assets may be served from any nearby service instance (since those base assets may be served without requiring cross-request state being maintained), while the client-specific input data is being processed by a stateful service instance that changes, if at all, only slowly over time due to the stickiness of the service that is being created by the client-specific data. Other splits of rendering and input tasks can be found in for further reading. When it comes to the service instances themselves, those may be instantiated on-demand, e.g., driven by network or client demand metrics, while resources may also be released, e.g., after an idle timeout, to free up resources for other services. Depending on the utilized node technologies, the lifetime of such "function as a service" may range from many minutes down to millisecond scale. Therefore, computing resources across participating edges exhibit a distributed (in terms of locations) as well as dynamic (in terms of resource availability) nature. In order to achieve a satisfying service quality to end users, a service request will need to be sent to and served by an edge with sufficient computing resource and a good network path.

Example 2: Computing-aware Intelligent Transportation For the convenience and safety of transportation, more video capture devices need to be deployed as urban infrastructure, and the better video quality is also required to facilitate the content analysis. Therefore, the transmission capacity of the network will need to be further increased, and the collected video data need to be further processed by edge nodes for edge computing, such as for pedestrian face recognition, vehicle tracking, and road accident prediction. This, in turn, also impacts the requirements for the video processing capacity of computing nodes and network capacity of network nodes in terms of network bandwidth and delay. In auxiliary driving scenarios, to help overcome a non-line-of-sight problem due to blind spot (or obstacles) and an abrupt collision problem, the edge node can collect comprehensive road and traffic information around the vehicles' locations and perform data processing. Then the vehicles with high security risk can be warned accordingly in advance and provided with safe manuveur guide (e.g., left-lane change, right-lane change, speed reduction, and braking) . This can improve driving safety in complicated road conditions, such as at intersections and on highways, through the help from the edge node having a wider view. This scenario is also called "Extended Electronic Horizon" , because the vehicles can extend their view range by exchanging their physical view information from their onboard camera and LiDAR with an adjacent edge node or other vehicles via Vehicle-to-Everything (V2X) communications. For instance, video images captured by an onboard camera in each vehicle are transmitted to the nearest edge node for processing. The notion of sending the request to the "nearest" edge node is important for being able to collate the video information of "nearby" vehicles, using relative location information among the vehicles. Furthermore, data privacy may lead to a requirement to process the data by an edge node (or an adjacent vehicle as a cluster node ) as close to the source as possible to limit the data's spread across many network components in the network. Nevertheless, load at specificly "closest" nodes may greatly vary, leading to the possibility for the closest edge node becoming overloaded. This may lead to a higher response time and therefore a delay in responding to the auxiliary driving request. As a result, there will be road traffic delays or even vehicle accidents in the road networks. Thus, the selection of a "closest" node should consider the node's current workload and the network condition from the source vehicle to the "closest" edge node. Hence, in such cases, delay-insensitive services such as in-vehicle infotainment (e.g., online music playing, online video watching, and navigation service) should be dispatched to other lightly-loaded edge nodes instead of local edge nodes even though the lightly-loaded edge nodes are a little far away from the service user vehicle. On the other hand, delay-sensitive services are preferentially processed locally to ensure the service availability, Qualty of Service (QoS), and Quality of Experience (QoE). Thus, according to delay requirements of services, the selection of appropriate edge nodes should be done. Also, since the vehicles keep moving along the roadways, the migration of contexts for the service user vehicles should be smoothly and proactively between the current edge node and the next edge node in an undisrupted way, considering the vehicles' service requirements. In video recognition scenarios, as both the number of waiting people and that of vehicles increase, more computing resources are needed to process the video contents. Traffic congestion and weekend personnel flows from a city's edge to the city's center are huge. Thus, an efficient network and computing capacity scheduling is also required for scalable services according to the number of users. Those would cause the overload of the nearest edge sites (or edge nodes) to become much severer if there is no extra method used. Therefore, some of the service request flows might be steered to other appropriate edge sites (or edge nodes) rather than simply the nearest one.

Example 3: Computing-aware Digital Twin A number of industry associations, such as the Industrial Digital Twin Association or the Digital Twin Consortium (https://www.digitaltwinconsortium.org/), have been founded to promote the concept of the Digital Twin (DT) for a number of use case areas, such as smart cities, transportation, industrial control, among others. The core concept of the DT is the "administrative shell" , which serves as a digital representation of the information and technical functionality pertaining to the "assets" (such as an industrial machinery, a transportation vehicle, an object in a smart city or others) that is intended to be managed, controlled, and actuated. As an example for industrial control, the programmable logic controller (PLC) may be virtualized and the functionality aggregated across a number of physical assets into a single administrative shell for the purpose of managing those assets. PLCs may be virtualized in order to move the PLC capabilities from the physical assets to the edge cloud. Several PLC instances may exist to enable load balancing and fail-over capabilities, while also enabling physical mobility of the asset and the connection to a suitable "nearby" PLC instance. With this, traffic dynamicity may be similar to that observed in the connected car scenario in the previous subsection. Crucial here is high availability and bounded latency since a failure of the (overall) PLC functionality may lead to a production line stop, while boundary violations of the latency may lead to loosing synchronization with other processes and, ultimately, to production faults, tool failures or similar. Particular attention in Digital Twin scenarios is given to the problem of data storage. Here, decentralization, not only driven by the scenario (such as outlined in the connected car scenario for cases of localized reasoning over data originating from driving vehicles) but also through proposed platform solutions, such as those in , plays an important role. With decentralization, endpoint relations between client and (storage) service instances may frequently change as a result.

Example 4: Computing-aware SD-WAN SD-WAN provides organizations or enterprises with centralized control over multiple sites which are network endpoints including branch offices, headquarters, data centers, clouds, and more. A enterprise may deploy their services and applications in different locations to achieve optimal performance. The traffic sent by a host will take the shortest WAN path to the closest server. However, the closet server may not be the best choice with lowest cost of network and computing resources for the host. If the path computation element can consider the computing dimension information in path computation, the best path with lowest cost can be provided. The computing related information can be the number of vCPUs of the VM running the application/services, CPU utilization rate, usage of memory, etc. The SD-WAN can be aware of the computing resource of applications deployed in the multiple sites and can perform the routing policy according to the information is defined as the computing-aware SD-WAN. Many enterprises are performing the cloud migration to migrate the applications from data centers to the clouds, including public, private, and hybrid clouds. The clouds resources can be from the same provider or multiple cloud providers which have some benefits including disaster recovery, load balancing, avoiding vendor lock-in. In such cloudification deployments SD-WAN provides enterprises with centralized control over Customer-Premises Equipments(CPEs) in branch offices and the cloudified CPEs(vCPEs) in the clouds. The CPEs connect the clients in branch offices and the application servers in clouds. The same application server in different clouds is called an application instance. Different application instances have different computing resource. SD-WAN is aware of the computing resource of applications deployed in the clouds by vCPEs, and selects the application instance for the client to visit according to computing power and the network state of WAN. Additionally, in order to provide cost-effective solutions, the SD-WAN may also consider cost, e.g., in terms of energy prices incurred or energy source used, when selecting a specific application instance over another. For this, suitable metric information would need to be exposed, e.g., by the cloud provider, in terms of utilized energy or incurred energy costs per computing resource. below illustrates Computing-aware SD-WAN for Enterprise Cloudification.

Illustration of Computing-aware SD-WAN for Enterprise Cloudification

The current computing load status of the application APP1 in cloud1 and cloud2 is as follows: each application uses 6 vCPUs. The load of application in cloud1 is 50%. The load of application in cloud2 is 20%. The computing resource of APP1 are collected by vCPE1 and vCPE2 respectively. Client1 and Client2 are visiting APP1 in cloud1. WAN1 and WAN2 have the same network states. Considering lightly loaded application SD-WAN selects APP1 in cloud2 for the client3 in branch office. The traffic of client3 follows the path: Client3 -> CPE -> WAN1 -> Cloud2 vCPE1 -> Cloud2 APP1

Example 5: Computing-aware Distributed AI Training and Inference Artificial Intelligence (AI) large model refers to models that are characterized by their large size, high complexity, and high computational requirements. AI large models have become increasingly important in various fields, such as natural language processing for text classification, computer vision for image classification and object detection, and speech recognition. AI large model contains two key phases: training and inference. Training refers to the process of developing an AI model by feeding it with large amounts of data and optimizing it to learn and improve its performance. On the other hand, inference is the process of using the trained AI model to make predictions or decisions based on new input data.

Distributed AI Inference With the fast development of AI large language models, more lightweight models can be deployed at edge sites. shows the potential deployment of this case. AI inference contains two major steps, prefilling and decoding. Prefilling processes a user’s prompt to generate the first token of the response in one step. Following it, decoding sequentially generates subsequent tokens step-by-step until the termination token. These stages consume much computing resource. Important metrics for AI inference are processor cores which transform prompts to tokens, and memory resources which are used to store key-values and cache tokens. the generation and processing of tokens judge the service capability of an AI inference system. Single site deployment of the prefilling and decoding might not provide enough resources when there are many clients sending requests (prompts) to access AI inference service. More generally, we also see the use of cost information, specifically on the cost for energy expended on AI inferencing of the overall provided AI-based service, as a possible criteria for steering traffic. Here, we envision (AI) service tiers being exposed to end users, allowing to prioritize, e.g., 'greener energy costs' as a key criteria for service fulfilment. For this, the system would employ metric information on, e.g., utilized energy mix at the inferencing sites and costs for energy to prioritize a 'greener' site over another, while providing similar response times.

Illustration of Computing-aware AI large model inference

Distributed AI Training Although large language models are nowadays confined to be trained with very large centers with computational, often GPU-based, resources, platforms for federated or distributed training are being positioned, specifically when employing edge computing resources. While those approaches apply their own (collective) communication approach to steer the training and gradient data towards the various (often edge) computing sites, we also see a case for CATS traffic steering here. For this, the training clusters themselves may be multi-site, i.e., combining resources from more than one site, but acting as service instances in a CATS sense, i.e., providing the respective training round as a service to the overall distributed/federated learning platform. One (cluster) site can be selected over another based on compute, network but also cost metrics, or a combination thereof. For instance, training may be constrained based on the network resources to ensure timely delivery of the required training and gradient information to the cluster site, while also computational load may be considered, particularly when the cluster sites are multi-homed, thus hosting more than one application and therefore become (temporarily) overloaded. But equally to our inferencing use case in the previous section, the overall training service may also be constrained by cost, specifically energy aspects, e.g., when positioning the service utilizing the trained model is advertising its 'green' credentials to the end users. For this, costs based on energy pricing (over time) as well as the energy mix may be considered. One could foresee, for instance, the coupling of surplus energy in renewable energy resources to a cost metric upon which traffic is steered preferably to those cluster sites that are merely consuming surplus and not grid energy. Storage is also necessary for performing distributed/federated learning due to several key reasons. Firstly, it is needed to store model checkpoints produced throughout the training process, allowing for progress tracking and recovery in case of interruptions. Additionally, storage is used to keep samples of the dataset used to train the model, which often come from distributed sensors such as cameras, microphones, etc. Furthermore, storage is required to hold the models themselves, which can be very large and complex. Knowing the storage performance metrics is also important. For instance, understanding the I/O transfer rate of the storage helps in determining the latency of accessing data from disk. Additionally, knowing the size of the storage is relevant to understand how many model checkpoints can be stored or the maximum size of the model that can be locally stored.

Requirements In the following, we outline the requirements for the CATS system to overcome the observed problems in the realization of the use cases above.

Support dynamic and effective selection among multiple service instances The basic requirement of CATS is to support the dynamic access to different service instances residing in multiple computing sites and then being aware of their status, which is also the fundamental model to enable the traffic steering and to further optimize the network and computing services. A unique service identifier is used by all the service instances for a specific service no matter which edge site an instance may attach to. The mapping of this service identifier to a network locator is basic to steer traffic to any of the service instances deployed in various edge sites. Moreover, according to CATS use cases, some applications require E2E low latency, which warrants a quick mapping of the service identifier to the network locator. This leads to naturally the in-band methods, involving the consideration of using metrics that are oriented towards compute capabilities and resources, and their correlation with services. Therefore, a desirable system R1: MUST provide a discovery and resolving method for the mapping of a service identifier to a specific address. R2: MUST provide a method to determine the availability of a service instance.

Support Agreement on Metric Representation and Definition Computing metrics can have many different semantics, particularly for being service-specific. Even the notion of a "computing load" metric could be represented in many different ways. Such representation may entail information on the semantics of the metric or it may be purely one or more semantic-free numerals. Agreement of the chosen representation among all service and network elements participating in the service instance selection decision is important. Therefore, a desirable system R3: MUST agree on using metrics that are oriented towards compute capabilities and resources and their representation among service instaces in the participating edges. To better understand the meaning of different metrics and to better support appropriate use of metrics, R4: A model of the compute and network resources MUST be defined. Such a model MUST characterize how metrics are abstracted out from the compute and network resources. We refer to this model as the Resource Model. R5: The Resource Model MUST be implementable in an interoperable manner. That is, independent implementations of the Resource Model must be interoperable. R6: The Resource Model MUST be executable in a scalable manner. That is, an agent implementing the Resource Model MUST be able to execute it at the required time scale and at an affordable cost (e.g., memory footprint, energy, etc.). R7: The Resource Model MUST be useful. That is, the metrics that an agent can obtain by executing the Resource Model must be useful to make node and path selection decisions. We recognize that different network nodes, e.g., routers, switches, etc., may have diversified capabilities even in the same routing domain, let alone in different administrative domains and from different vendors. Therefore, to work properly in a CATS system, R8: There MUST set up metric information that can be understood by CATS components. For metrics that CATS components do not understand or support, CATS components will ignore them. R9: The computing metrics in CATS MUST be simple, that is distributing metrics and selecting path based on these metrics will not cause routing loops and route oscillation.

Use of CATS Metrics Network path costs in the current routing system usually do not change very frequently. Network traffic engineering metrics (such as available bandwidth) may change more frequently as traffic demands fluctuate, but distribution of these changes is normally damped so that only significant changes cause routing protocol messages. However, metrics that are oriented towards compute capabilities and resources in general can be highly dynamic, e.g., changing rapidly with the number of sessions, the CPU/GPU utilization and the memory consumption, etc. It has to be determined at what interval or based on what events such information needs to be distributed. Overly frequent distribution with more accurate synchronization may result in unnecessary overhead in terms of signaling. Moreover, depending on the service related decision logic, one or more metrics need to be conveyed in a CATS domain. The problem to be addressed here may be the frequency of such conveyance, thanks to the comprehensive load that a signaling process may add to the overall network traffic. While existing routing protocols may serve as a baseline for signaling metrics, other means to convey the metrics can equally be considered and even be realized. Specifically, a desirable system R10: MUST provide mechanisms for metric collection. R11: MUST declare the entity that collect metrics. Collecting metrics from all of the services instances may incur much overhead for the decision maker, and thus hierarchical metric collection is needed. That is, R12: SHOULD provide mechanisms to aggregate the metrics. CATS components do not need to be aware of how metrics are collected behind the aggregator. The decision point may not be directly connected with service instances or metric collectors, therefore， R13: MUST provide mechanisms to distribute the metrics. The update frequency of the computing metrics may be various. Some of the metrics may be more dynamic, and some are relatively static. Accordingly, different distribution methods may be chosen with respect to different update frequencies of relevant metrics. Therefore, R14: MUST be clear of the update frequency of CATS metrics and its corresponding distribution method. Sometimes, a metric that is chosen is not accurate for service instance selection, in such case, a desirable system R15: SHOULD provide mechanisms to assess selection accuracy and re-select metrics if the selection result is not accurate.

Support Instance Affinity In the CATS system, a service may be provided by one or more service instances that would be deployed at different locations in the network. Each instance provides equivalent service functionality to their respective clients. The decision logic of the instance selection are subject to the normal packet level communication and packets are forwarded based on the operating status of both network and computing resources. This resource status will likely change over time, leading to individual packets potentially being sent to different network locations, possibly segmenting individual service transactions and breaking service-level semantics. Moreover, when a client moves, the access point might change and successively lead to the same result of the change of service instance. If execution changes from one (e.g., virtualized) service instance to another, state/context needs transfer to another. Such required transfer of state/context makes it desirable to have instance affinity as the default, removing the need for explicit context transfer, while also supporting an explicit state/context transfer (e.g., when metrics change significantly). So in those situations: R16: Instance affinity MUST be maintained when the transaction is stateful. The nature of this affinity is highly dependent on the nature of the service, which could be seen as a 'instance affinity' to represent the relationship. The minimal affinity of a single request represents a stateless service, where each service request may be responded to without any state being held at the service instance for fulfilling the request. Providing any necessary information/state in-band as part of the service request, e.g., in the form of a multi-form body in an HTTP request or through the URL provided as part of the request, is one way to achieve such stateless nature. Alternatively, the affinity to a particular service instance may span more than one request, as in the AR/VR use case, where previous client input is needed to render subsequent frames. However, a client, e.g., a mobile UE, may have many applications running. If all, or majority, of the applications request the CATS- based services, then the runtime states that need to be created and accordingly maintained would require high granularity. In the extreme scenario, this granular requirement could reach the level of per-UE, per-APP, and per-(sub)flow with regard to a service instance. Evidently, these fine-granular runtime states can potentially place a heavy burden on network devices if they have to dynamically create and maintain them. On the other hand, it is not appropriate either to place the state-keeping task on clients themselves. Besides, there might be the case that UE moves to a new (access) network or the service instance is migrated to another cloud, which cause the unreachable or inconvenient of the original service instance. So the UE and service instance mobility also need to be considered. Therefore, a desirable system R17: Instance affinity MUST be maintained for service requests or transactions that belong to the same flow. R18: MUST avoid keeping fine runtime-state granularity in network nodes for providing instance affinity. For example, as mentioned above, maintaining per-flow states for a specific APP. R19: MUST provide mechanisms to minimize client side states in order to achieve the instance affinity. R20: SHOULD support the UE and service instance mobility.

Preserve Communication Confidentiality Exposing the information of computing resources to the network may lead to the leakage of computing domain and application privacy. In order to prevent it, it need to consider the methods to process the sensitive information related to computing domain. For instance, using general anonymous methods, including hiding the key information representing the identification of devices, or using an index to represent the service level of computing resources, or using customized information exposure strategies according to specific application requirements or network scheduling requirements. At the same time, when anonymity is achieved, it is also necessary to consider whether the computing information exposed in the network can help make full use of traffic steering. Therefore, a CATS system R21: MUST preserve the confidentiality of the communication relation between user and service provider by minimizing the exposure of user-relevant information according to user needs.

Correlation between Use Cases and Requirements A table is presented in this section to better illustrate the correlation between CATS use cases and requirements, 'X' is for marking that the requirement can be derived from the corresponding use case.

Mapping between CATS Use Cases and Requirements

Security Considerations CATS decision making process is deeply related to computing and network status as well as some service information. Some security issues need to be considered when designing CATS system. Service data sometimes needs to be moved among different edge sites to maintain service consistency and availability. Therefore: R22: service data MUST be protected from interception. The act of making compute requests may reveal the nature of user's activities, so that: R23: the nature of user's activities SHOULD be hidden as much as possible. The behavior of the network can be adversely affected by modifying or interfering with advertisements of computing resource availability. Such attacks could deprive users' of the services they desires, and might be used to divert traffic to interception points. Therefore, R24: secure advertisements are REQUIRED to prevent rogue nodes from participating in the network. Compromised or malicious computing resources will bring threats to network and services of users, such as data of services maybe stolen, output of computing services may have deviations and other computing resources are attacked by the compromised resources that collaborate with them for computation. Therefore, R25: When making service decisions, the security status of computing resources SHOULD be taken into consideration.

IANA Considerations This document makes no requests for IANA action.

References Normative References Applications using the Internet already have access to some topology information of Internet Service Provider (ISP) networks. For example, views to Internet routing tables at Looking Glass servers are available and can be practically downloaded to many network application clients. What is missing is knowledge of the underlying network topologies from the point of view of ISPs. In other words, what an ISP prefers in terms of traffic optimization -- and a way to distribute it. The Application-Layer Traffic Optimization (ALTO) services defined in this document provide network information (e.g., basic network location structure and preferences of network paths) with the goal of modifying network resource consumption patterns while maintaining or improving application performance. The basic information of ALTO is based on abstract maps of a network. These maps provide a simplified view, yet enough information about a network for applications to effectively utilize them. Additional services are built on top of the maps. This document describes a protocol implementing the ALTO services. Although the ALTO services would primarily be provided by ISPs, other entities, such as content service providers, could also provide ALTO services. Applications that could use the ALTO services are those that have a choice to which end points to connect. Examples of such applications are peer-to-peer (P2P) and content delivery networks. 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 Telefonica Nokia Bell Labs Qualcomm Europe Benocs Unicamp Service providers are starting to deploy computing capabilities across the network for hosting applications such as AR/VR, vehicle networks, IoT, and AI training, among others. In these distributed computing environments, knowledge about computing and communication resources is necessary to determine the proper deployment location of each application. This document proposes an initial approach towards the use of ALTO to expose such information to the applications and assist the selection of their deployment locations. 3GPP Proceedings of 14th International Conference on ITS Telecommunications (ITST) Industry4.0 Huawei Technologies China Mobile Orange Telefonica Juniper Networks, Inc. This document describes a framework for Computing-Aware Traffic Steering (CATS). Specifically, the document identifies a set of CATS components, describes their interactions, and provides illustrative workflows of the control and data planes.

Appendix A This section presents several attempts to apply CATS solutions for improving service performance in different use cases. It is a temporary and procedural section which might be deleted or merged in future updates. The major reason is to help facilitate the discussion and definition of CATS metrics and solidify CATS framework and CATS solutions.

CATS for Content Delivery Network(CDN) CDN is mature enough to deploy contents like high resolution videos near clients, so as to provide good performance. However, when existing CDN servers can not guarantee the quality of service, for example, there is not enough query per second(QPS) offering capabilities, CATS can help relieve the problem and improve the overall performance through better load balancing. Two deployment methods of CATS are tested in two different provinces of China, i.e. distributed solution and centralized solution. Some preliminary results show that CATS can guarantee the service performance at client side when there are large number of concurrent sessions coming, compared to existing CDN scheduling solutions. Key performance indicators include the end-to-end scheduling delay, average computing capabities after load balancing(measured as queries per second). below illustrates the deployment of CATS solution in ten cities of Henan, China. Two ingress nodes are deployed in Zhengzhou, which is the provincial capital of Henan. All egress nodes are deployed in the other nine cities, with each egress node settled in only one city respectively. Client terminals are deployed near ingress nodes in Zhengzhou. The provincial networking can test the performance gains of CATS solutions over 100 kilometers.

Deployment of CATS in ten cities of Henan, China

below illustrates the deployment of CATS solution in three cities of Jiangsu, China. The ingress node is deployed in Wuxi, while the other two egress nodes are deployed in Xuzhou and Suzhou, respectively. Client devices like laptops and the centralized controller are deployed near the ingress node in Wuxi, since Wuxi owns the largest computing capabilities inside the city and can support over 200,000 connections per second at its peak value. CATS can help alleviate the stress of load balancing at peak hours.

Deployment of CATS in three cities of Jiangsu, China

From the experiments in CDN, benefits of CATS can be summarized as follows. CATS can adapt to network quality degradation more timely than traditional approaches. The frequency of DNS request for available service instances is set to be 600 seconds normally, which is a bit too long when the network quality can not be guaranteed. In a CATS system, the metric update and distribution frequency is set to be 15 seconds in this case, which is the same as the normal refresh frequency of BGP update. Therefore, after the first DNS request for a service instance, CATS will alternatively select other instances no later than 15 seconds if the current service instance do not work well or the quality of path towards this instance drops.

CATS for MIGU Cloud Rendering MIGU is the digital content subsidiary of China Mobile, offering various digital and interactive applications which are enriched by 5G, cloud rendering, and AI. Cloud rendering needs a lot of compute resources to be deployed at edge sites, in order to satisfy the real-time modeling requirements. In cooperation with China Mobile Zhejiang Co. Ltd, MIGU has deployed its cloud rendering applications at various edge sites in Zhejiang, in order to test how CATS solution can improve the overall performance of image rendering. Key performance indicators include the resolution and sharpness of images or videos, and processing delay. below illustrate the deployment of MIGU cloud rendering applications in Zhejiang. Three edge sites are deployed in Wenzhou, Ningbo, and Jiaxing, respectively. In each site, there are some servers for processing rendering services and some corresponding management nodes. One CATS egress node is deployed outside each site for service instance selection. There is a MIGU cloud platform which collects the compute metrics from three different sites, and then it passes these information to the central controller and the controller will synchronize these information to the ingress node which is deployed in Hanzhou, near the client.

Deployment of CATS for MIGU App Performance Improvement in Zhejiang, China

CATS for High-speed Internet of Vehicles In high-speed Internet of vehicles (IoV), high-speed vehicles, such as cars, trains, subways, etc., need to communicate with other vehicles, infrastructure or cloud service through network to run applications carried by vehicles. These applications can be divided into two types. One type of applications will affect the driving, such as autonomous driving, remote control, and intelligent driving services. They have extremely high requirements on network delay, and the console needs to make quick judgments and responses. Otherwise, as the vehicle travels quickly, a brief misoperation may lead to extremely serious traffic accidents. And they have extremely high requirements for service switching efficiency. The coverage of a base station is limited, and the capabilities of different service sites are also different. Vehicle movement will inevitably cause switching of access station and service site. The delay and service changes caused by switching also directly affect the experience and safety of driving. Another type of applications is not related to driving, such as voice communications, streaming media, and other entertainment services. They do not have strict requirements on real-time performance and service switching efficiency, but may requires higher computing capability. Due to the complex requirements of high-speed IoV on computing capability, an efficient way is needed to jointly schedule computing and network resources.. The hybrid CATS scheme combines both the characteristics of centralized and distributed schemes. In hybrid CATS scheme, the awareness and advertisement of computing status are performed by a centralized orchestration system, service selection and path computation are performed by network devices. As shown in , the centralized computing status advertisement can reduce the deployment hurdle of CATS.

Deployment of CATS for Intelligent Transportation in Hebei, China

The hybrid CATS scheme based high-speed IoV solution has been deployed and validated for the first time in Hebei, China by China Unicom. The computing and network status were comprehensively used for service selection and path computation, which provided high quality computing service with the optimal service site and optimal forwarding path for vehicle terminal applications. Distributed routing mode provides real-time optimization and fast switching capabilities for delay-sensitive applications, such as the autonomous driving. Some non-delay sensitive applications, such as online media and entertainment, used centralized routing decision mode to achieve global resource scheduling.

Acknowledgements The authors would like to thank Adrian Farrel, Peng Liu, Luigi Iannone, Christian Jacquenet and Yuexia Fu for their valuable suggestions to this document. The authors would like to thank Yizhou Li for her early IETF work of Compute First Network (CFN) and Dynamic Anycast (Dyncast) which inspired the CATS work.

Contributors The following people have substantially contributed to this document: Huawei Technologies

liyizhou@huawei.com

Huawei Technologies

dirk.trossen@huawei.com

Orange

mohamed.boucadair@orange.com

pjw7904@rjt.edu

philip.eardley@googlemail.com

China Mobile

tianjijiang@chinamobile.com

Deutsche Telekom

Markus.Amend@telekom.de

ZTE

huang.guangping@zte.com.cn

ZTE

yuan.dongyu@zte.com.cn

China Unicom

yixx3@chinaunicom.cn

China Unicom

yixx3@chinaunicom.cn

Qualcomm Europe, Inc.

jros@qti.qualcomm.com

Sungkyunkwan University

pauljeong@skku.edu