Casablanca Platform Enhancements to Support 5G Use Case (Network Deployment, Slicing, Network Optimization and Automation Framework) 5G Use Case Team

Contributors: Supporting Companies

High Level Objectives for Casablanca: Deployment of the hybrid 5G Radio Network (PNFs & VNFs) Complete PNF Support Platform Enhancements to Deploy Edge Virtual Radio Network Functions (e.g. CU) Creation & Management of Radio Network Slice Subnet within the deployed network Lifecycle management of Radio Network Slice Subnet Instances Optimization of the deployed 5G network and the slices Near-real Time Data Collection (e.g. streaming) Edge Analytics to Support 5G Network Optimization OOF enhancements for optimal placement of edge resources

A: 5G Radio Network – Architecture & Casablanca Focus ONAP Management - Design ONAP Management - Runtime 5G Application Ecosystem RAN Network Elements Core Network Elements Macro Radio & Small Cell Antennas CU-CP RU DU CU-UP Centralized Cloud Internet UE/CPE Back Haul External Content Front Haul Front Haul II Edge Cloud RU – Remote Radio Unit DU – Distributed Unit CU-UP – Centralized Unit - User Plane CU-CP – Centralized Control Plane

A: 5G Radio Network Deployment Requirements Description Disaggregated 5G RAN may include PNFs and VNFs, in which case cloud infrastructure deployment at the edge is required. Beijing implemented the first phase of PNF discovery and instantiation. Our goal for Casablanca is expand on that work, include VNF deployment at the edge, and fully integrated lifecycle management. Key enhancements needed are: Support full Application level Configuration (+Ansible), allow various mobile network elements to be controlled from same controller persona created from CC-SDK Add Lifecycle management functions to controller persona Support an integrated configuration design tool in SDC that can be used with any controller persona (next gen CDT) Add support for PNF Software Management and Change management Edge Cloud Support Add needed support for deploying Mobility Virtual Network Elements (e.g. CU) at the Edge locations Further automation of PnP Discovery for PNF Rationale Support for deploying and managing 5G mobility network is critically important for most ONAP members. Impacted ONAP components SDC, SO, CC-SDK (SDN-C, APP-C), AAI, DCAE, Multi-Cloud, OOF, (V/P)NF-SDK Participating Companies AT&T Amdocs China Mobile Ciena Cisco Ericsson Huawei Intel Nokia VMWare Others For details regarding the requirements, please see https://wiki.onap.org/display/DW/Missing+Platform+capabilities

B: Casablanca Scope – Radio Network Slice Management ONAP Management - Design ONAP Management - Runtime Mobile Slice Services (e.g. eMBB, Massive IoT, Mission-critical IoT) Network Slice Instances (NSIs) Radio Network Slice Subnet Instances (NSSIs) Transport Network Slice Subnet Instances (NSSIs) Core Network Slice Subnet Instances (NSSIs) 5G Application Ecosystem RAN Abstraction Model Transport Abstraction Model Core Abstraction Model Radio Antennas Core Network Elements CU-CP CU-UP Centralized Cloud Internet RU DU Back Haul Front Haul Front Haul II External Content Edge Cloud UE/CPE RU – Remote Radio Unit DU – Distributed Unit (5G Base Unit) CU-UP – Centralized Unit, User Plane CU-CP – Centralized Unit, Control Plane

B: Radio Network Slice Management – ONAP Requirements Description Many providers are interested in using ONAP for slice management and extending the cloud notion of sharing network/compute/storage to sharing network functions and network slice subnet instances . 3GPP standards are still evolving and early versions are not expected to be available before June 2018. However, we would like to demonstrate these concepts in ONAP initially using a simple slice definition, creation, and management for 5G Radio Network. Radio Access Network slice subnet instances supported and demonstrated in Casablanca will be limited to simple examples such as eMBB (enhance Mobile Broad Band). Key enhancements needed are: Enhance SDC to model & define network slice subnets for Radio network Enhance SO Instantiate and lifecycle manage network slice subnet instances Enhance A&AI to inventory and store state of slice segments Enhance Multi-Cloud and A&AI to support Guarantee (hard isolation), Burstable (soft isolation) and Best Effort (no isolation) QoS classes for Edge Clouds Integrated design (e.g. configuration and service designs) Support Complex Service Modeling (e.g. service composition, nested allotted resources) Define and instantiate monitoring and management of network slice subnet instances, network slice instances Rationale Typical carrier deployments will be complex enough to need support for nested / complex services, correlation across multiple elements - hence, these are critical functional requirements for a carrier-grade ONAP Impacted ONAP components SDC, SO, CC-SDK (SDN-C, APP-C), AAI, Multi-Cloud, DCAE Participating Companies AT&T Amdocs China Mobile Ciena Cisco Ericsson Huawei Intel Nokia

Impacted ONAP components Participating Companies C: Near-Real Time Edge Analytics & Network Optimization ONAP Requirements Description Edge cloud infrastructure (ECI) analytics is required for near-real-time correlation between various Events (e.g. Number of VM Power On/Off); Alerts (e.g. Cloud region CPU usage exceeds threshold) and Faults (e.g. loss of redundancy from a Host NIC perspective). Edge cloud near-real-time streaming infrastructure resource (CPU, Memory, Network etc.) utilization metrics are needed by OOF for inter-cloud resource optimization Key Enhancements needed Enhance ONAP SDC to support new ECI Models Enhance ONAP Policy to support new ECI Policies Enhance ONAP Policy to synchronize ECI Policies with the ECI Analytics through ONAP Multi-Cloud Enhance ONAP Multi-Cloud to translate from vendor-specific to vendor-agnostic ECI Policies Enhance ONAP Multi-Cloud to translate from vendor-specific to vendor-agnostic Resource Utilization Metrics & Analytics and asynchronously communicate these over DMaaP Rationale Near-real time decision making is critical for proactive self-healing and inter-cloud resource optimization Important to minimize effects of WAN latency and bandwidth for near-real time decision making -- ONAP Components are centralized whereas Edge Clouds are Distributed Impacted ONAP components SDC (including Policy Design), Multi-cloud, Policy, OOF Participating Companies AT&T Amdocs China Mobile Ciena Cisco Ericsson Huawei Intel Nokia VMware

C: ONAP Optimization Framework Enhancements Description Many services in ONAP require optimization functionality during the lifecycle of the service components. Service design, deployment, management, scheduling, and dynamic reconfiguration The ONAP Optimization Framework provides declarative, model- and policy- driven system to support above needs. We would like to demonstrate OOF functionality using applications such as SON (PCI assignments, balancing network load), and dynamic workload aware placement and migration of edge services for 5G. Key enhancements needed are: Enhance A&AI to inventory and store 5G related data Enhance Multi-cloud to provide streaming updates (near-real time and/or feed of statistical aggregates) Enhance OOF to provide near-real time solutions and evaluate it for large clouds Enhance Policy to support complex service modeling (e.g. complex constraints, nested resources and related policies) Enhance SDC to model & define constraint and optimization policies Enhance SO to support workflows related to management of 5G related network entities Rationale Optimization needs for typical carrier deployments will involve large, distributed data centers, and will involve complex rules and constraints governing how multiple network elements are instantiated and managed. Hence, these are critical functional requirements for a carrier-grade ONAP. Specifically, within the 5G domain, efficient allocation of resources and dynamic, load- and cost-aware reconfigurations are valuable. Impacts Impacted ONAP components are: A&AI, Multi-cloud, Policy, OOF, SDC, SO Companies Participating companies are: AT&T, Cisco, Intel, Netcracker, VMWare

Thank You!