alto Working Group TK.Yu, H.Yang, ZJ.Sun, QY.Yao Internet-Draft Beijing University of Posts and Telecommunications Intended status: Standards Track Y.Zhao,S.Liu,YB.Li Expires: 06 March 2024 China Mobile Research Istitute 06 September 2023 Architecture of Metro Computing Power Optical Network draft-yu-alto-arch-metro-computing-optical-network-00 Abstract This document describes the architecture of metro computing power optical network. Status of This Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at https://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." This Internet-Draft will expire on 6 September 2023. Copyright Notice Copyright (c) 2023 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/ license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License. Yu, et al. Expires 6 September 2023 [Page 1] Internet-Draft Architecture of Metro Computing Power Optical Network September 2023 Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 2. Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.1. Network Resource Acquirement . . . . . . . . . . . . . . . 3 3. The architecture of Metro Optical Computing Power Network . . . 4 3.1. Edge computing power management platform . . . . . . . . . 4 3.2. Traffic scheduling management platform . . . . . . . . . . 4 4. Manageability Considerations . . . . . . . . . . . . . . . . 4 5. Security Considerations . . . . . . . . . . . . . . . . . . . 4 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 4 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 5 7.1. Normative References . . . . . . . . . . . . . . . . . . 5 7.2. Informative References . . . . . . . . . . . . . . . . . 5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . 5 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 5 1. Introduction The emergence of novel computing services, such as large-scale models and virtual reality, has introduced considerable challenges to communication networks due to their substantial data transmission and computational demands. These challenges are particularly pronounced within metropolitan networks. Heterogeneous computing devices pose difficulties in uniform measurement scheduling, accurately pinpointing diverse user computing intents, and handling scattered datasets, thereby hindering effective information utilization. These issues underscore the necessity to establish real-time transmission connections with significant bandwidth capacity. The metropolitan optical network, grounded in Optical transport Network (OTN) devices, facilitates optical connections and configures a three-layer hierarchical topology spanning access, aggregation, and core metro nodes. This architecture offers a favorable framework for the judicious allocation of computating power to individual users in accordance with service requirements. Optical networks expand the scope of computing networks, thereby affording more expansive avenues for implementing flexible and efficient scheduling strategies, optimizing the utilization of computating resources. This flexibility supports the deployment of ultra-large capacity, ultra-low latency, and highly efficient computing services. The architecture of the metropolitan optical computing power network facilitates diverse applications, encompassing computing power network measurement, perception, traffic prediction and control, as well as anomaly detection. Businesses seeking computing services can harmonize computing power and network resources to offer an optimized user experience. This architectural framework synergizes computational prowess with the metro optical network, facilitating collaborative resource allocation at the network edge. Yu, et al. Expires 6 September 2023 [Page 2] Internet-Draft Architecture of Metro Computing Power Optical Network September 2023 1.1. Requirements Language 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 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here. 2. Scenarios With the prevalence of cloud services, enterprise services and other services, the architecture of computing power optical network has become the choice to solve supported services. The following scenarios provide some typical applications. 2.1. Network Resource Acquirement The network traffic scheduling platform receives computing service information from clients, acquires comprehensive user data, and forwards it to the edge computing management platform for demand analysis. Subsequently, the scheduling platform conducts bandwidth resource forecasting and traffic scheduling based on these requirements. 3. The architecture of Metro Computing Power Optical Network ---------------------------------------------------------------------------- | --------------------------- ---------------------------- | | | Edge computing power | | Traffic scheduling | | | Metro | management | | management | | | optical | ----------------------- | | ---------------------- | | | computing | | CP cooperative process| | | | Service priorities | | | | power | ----------------------- |..| -.--------------.----- | | | network | ----------------------- | | -.------- ----.----- | | | management | | CP intent mapping | | | | Traffic | | Routing | | | | | ----------------------- | | | forecast| | and | | | | | ----------------------- | | | based | | spectrum | | | | | | CP state perception | | | | on AI | |assingment| | | | | ----------------------- | | ---.----- -------.-- | | | | ----------------------- | | ---.---------------.-- | | | | | CP measurement | |..| | Service Interference | | | | | ----------------------- | | ---------------/--/--- | | | --------------------------- -----------------/--/------- | -------------------------------------------------------------/--/----------- - Reporting edge computing power service information / / Scheduling -----------------------------------------------------------/--/-- scheme | metro backbone | | Optical OE ------------------ OE | | layer / \ / | | resources / \ metro core / | | OE - OE----------- OE | | / / \ -- / / \ | | -- / \ / / \ | | / OE OE OE OE OE | | OE . . metro aggregation . | -----------------.----.-----.-------------.------.-------.------- - . . . . . . -----------------.----.-----.-------------.-------.-------.------ | . . . . . . | | Computing CP . . . . . | | power CP CP . . CP | | resources CP CP | | | ----------------------------------------------------------------- Fig.1 The architecture of metro optical computing power network. 3.1. Edge computing power management platform To accurately fulfill the computational demands of services, the Edge Computing Power Management Platform comprises four integral modules: Computing Power measurement, Computing Power State Perception, Computing Power Intention Mapping, and Computing Power Collaborative Processing. The Edge Controller invokes the AI engine for embedded deployment and algorithm adaptation. The platform interfaces with the Traffic Scheduling Management Platform to process incoming service characteristic data, ascertain customized computational power requirements, and subsequently relay this information to the Traffic Scheduling Module for unified service transmission. These processes are incremental and sequential. Initially, it involves computation and state perception, followed by intention analysis and mapping based on service characteristics, ultimately concluding with the querying of computational state resources for collaborative processing. Computational power encompasses a diverse array of computational infrastructure capabilities, including processing speed, memory capacity, cache size, energy consumption, and failure rates. This infrastructure encompasses a combination of various computing elements, such as single-core central processing units (CPUs), graphics processing units (GPUs), and network processors (NPUs). To accommodate the varied computational requirements in the realm of edge computing, this layer offers functionalities such as an algorithm library, computing application programming interfaces (APIs), and the identification of computing network resources.The state perception of computational power involves monitoring the remaining resources of computational infrastructure over time scales, mitigating resource shortages or underutilization in metropolitan areas to achieve load balancing. Computational intention mapping entails the analysis and mapping of computational service intentions. It utilizes deep learning models to precisely match network transmission resources with computing resources in accordance with service needs. Computational power collaborative processing refers to multi-point collaborative edge processing guided by service intent and network resource states. Yu, et al. Expires 6 September 2023 [Page 3] Internet-Draft Architecture of Metro Optical Computing Power Network September 2023 3.2. Traffic scheduling management platform The Traffic Scheduling Management Platform comprises several critical components, including the Edge Service Interface, Deep Learning-Based Traffic Prediction, Routing and Spectrum Resource Allocation, and Service Priority Matching. This platform plays a pivotal role in the efficient management of service requests. Service requests from clients are channeled to the Traffic Scheduling Management Platform via the service interface, where they are subsequently forwarded to the Edge Computing Management Platform for processing. The platform, leveraging predictions of computational traffic and the processing outcomes from the computing management platform, proceeds to ascertain service priorities. It then proceeds to manage routing, spectrum resource allocation, and computing power scheduling. The Edge Service Interfaces encompass modules for network resource virtualization, which analyze underlying services and employ service classification methodologies. Business characteristics and requirements are conveyed to the Edge Computing Power Management Platform for further processing. Deep Learning-Based Traffic Prediction encompasses models for burst traffic detection, service type and arrival time prediction, as well as predictions of required network bandwidth and computing power resources. This segment is responsible for forecasting computational power demands and offering corresponding guidance. The Routing and Spectrum Resource Allocation model oversees downstream resource allocation following the analysis of computational power processing outcomes, efficiently managing resources for the corresponding computational power services. The Service Priority Matching model integrates priority-related processing algorithms, including those for delay-sensitive scenarios and bandwidth fragmentation, to provide more finely-tuned computational power services for service delivery. 4. Manageability Considerations TBD 5. Security Considerations TBD 6. IANA Considerations This document requires no IANA actions. 7. References Yu, et al. Expires 6 September 2023 [Page 4] Internet-Draft Architecture of Metro Optical Computing Power Network September 2023 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017, . Acknowledgments TBD Authors' Addresses Tiankuo Yu Beijing University of Posts and Telecommunications Email: yutiankuo@bupt.edu.cn Hui Yang Beijing University of Posts and Telecommunications Email: yanghui@bupt.edu.cn Zhengjie Sun Beijing University of Posts and Telecommunications Email: sunzhengjie@bupt.edu.cn Qiuyan Yao Beijing University of Posts and Telecommunications Email: yqy89716@bupt.edu.cn Yang Zhao China Mobile Email: zhaoyangyjy@chinamobile.com Sheng Liu China Mobile Email: liushengwl@chinamobile.com Yunbo Li China Mobile Email: liyunbo@chinamobile.com Yu, et al. Expires 6 September 2023 [Page 5] Internet-Draft Architecture of Metro Optical Computing Power Network September 2023