5G Cloud Native from RAN to Core Christian Maciocco, Intel Shilpa - PowerPoint PPT Presentation

5G Cloud Native from RAN to Core Christian Maciocco, Intel Shilpa Talwar, Intel Saikrishna Edupuganti, Intel Muhammad (Asim) Jamshed, Intel 2020

Agenda Cloud Native Disaggregated Network Infrastructure • Transition to 5G • Near Real-Time RAN Information Controller & Services • Demo of Dual Mode 5G UPF •

Build Open 5G Network Infrastructure to Accelerate Edge Deployment Devices Edge Access Core Cloud Compute Network Network Data Center Video Healthcare Manufacturing Drivers for edge Smart Devices Energy Latency, Bandwidth Retail Security, connectivity Transportation Smart Robots & Cities Industrial Public Sector Mobile Core RIC Visual Cloud, Industrial IOT , Control Plane Smart city, v2x,, … CU Cloud Gaming Immersive Media DU Access & Core move closer to Mobile Core the edge(s) to process data Data Plane Radio Unit Disaggregated Core and RAN on high Media Analytics volume server / programmable devices 5G workloads and open solutions will offer insights for architecture and system partitioning challenges

Building on ONF Success Disaggregating a 4G/LTE Core Operators create Common spec. RFP & Reference Designs become “gold standards” for basis of RFPs Reference Platform Designs From open Impact source to Exemplar Deployments Solutions Trials Platforms deployment Open Source Components Charge Trigger Policy Function Home Offline HSS Charging (CTF) Subscriptio Charging DB Rules n Server Service Charge Data Function (HSS) (OFCS) Function (PCRF) (CDF) Mobility Service Packet Manageme Gateway Gateway SGX Key Store nt Entity Control Control Control (MME) (SGW-C) (PGW-C) SGX Billing Service Packet Gateway Gateway Access Internet Data Data User Data User Data Network (SGW-U) (PGW-U) Disaggregated SPGW SW deployed by DT/T-Mo Deutsche Telekom/T-Mobile Poland Production Deployment 4

Towards 5G SA - A Dual Mode 5G/LTE UPF 5G SBA N13 NSSF AUSF UDM N11 SMF (Service Base Architecture) N12 N8 N10 N22 PFCP PCF AF AMF SMF N5 N11 N7 N4 N15 N4 N4 UPF N2 N3 N6 Data N9 A-UPF I-UPF Network N6 UPF Slow Path (and Control) Data Network P4RT gRPC gRPC/P4RT DN N3/N9 N6 UPF Fast Path Data Network Di Disaggregated U UPF

UPF with One Slow Path, Fast Path Options SMF SMF SMF PFCP PFCP PFCP UPF Slow Path UPF Slow Path UPF Slow Path (and Control) (and Control) (and Control) P4RT gRPC P4RT gRPC P4RT gRPC gRPC P4RT P4RT gRPC HW Fast Path* SW Fast Path SW Fast Path HW Fast Path* (Tofino – P4) (DPDK based) (DPDK based) (Tofino Switch) Offload to accelerator Offload to host/NIC/FPGA (SmartNIC, FPGA, …) (for e.g. hQoS, DDN, large tables) * P4 Pipeline developed at ONF SW Fast Path Pros: HW Fast Path Pros: • Flexibility & support all features including • Aggregate throughput hQoS, DDN, DPI, FW Latency & jitter • • Support very large users’ table Limitations vs. SW Fast path: Use of platform features : DDP, DLB, SGX Need to offload to CPU/FPGA/SmartNIC to • • Limitations vs. HW Fast Path: support hQoS, DDN, DPI, FW • Aggregate throughput • Support for large number of users (flows DDN: Downlink Data Notification Higher latency & jitter in/out of TCAM create exception) • hQoS: Hierarchical QOS DPI: Deep Packet Inspection FW: Firewall A flexible 5G UPF architecture optimized for specific deployment, e.g. edge or Central Office DDP: Dynamic Device Personalization DLB: Dynamic Load Balancing SGX: Secure Enclave

Build UPF Processing Pipeline PFCP UPF Slow Path (and Control) Req. PFCP-ID Resp. PFCP-ID UPF Fast Path Packet PFCP PDR Parsing and Session BARs URRs FARs QERs PDR Packet In Metadata PDR Packet Out Acquisition Lookup PDR Apply instructions from the PDR N3 / N9 N6 / N9 …. PDR QER URR BAR Metadata Session + FAR Table Counter Counter ETH ETH hQoS & Table extraction PDR Table Post-QoS Pre-QoS Per- SRR IP IP Per-PDR PDR UD UD Keys : Keys : P P Keys : FAR-ID Meta data [ ] FAR BAR-ID F-SEID GTP GTP Executor: - UE IP address ... ... Values : - Src Interface Values: Forwards, Drops, - F-SEID - TEID Action Type, or Tunnels - PDR-ID - Dest IP Tunnel out type Vals : - FAR-ID Tunnel out Src IP - CTR-ID Tunnel out Dst IP - QFI Tunnel out TEID Tunnel out UDP port PFCP Session: PDR [ ], FAR [ ], BAR [ ], URR [ ], QER [ ], SRR [ ], … UPF supports dual-mode 5G and LTE Core PDR : Packet Detection Rule [ ] FAR : Forwarding Action Rule [ ] e.g. drop, forward, buffer, notify CP, duplicate, … 5G : UPF Interoperating with Spirent 5G Emulator and other emulators BAR : Buffering Action Rule, e.g. how much data to buffer and how to notify the CP 4G : Deployed on Aether’s edges QER : QoS Enforcement Rule [ ] -- Flow and service level marking URR: Usage Reporting Rule [ ] -- Generate reports to enable charging functionality

Cloud Native SW w/ Enhance Platform Awareness (EPA) (1/3) • CPU Core isolation & pinning • Huge Pages • Containers with multi-network interfaces & SR-IOV support in K8s Core pinning/affinity and isolation Huge pages § § CPU Manager for K8s § Automated CPU core mask for DPDK apps § Memory core 0 core 0 App Huge Page App Pin & Isolate Check TLB Cache Address core 1 (1 GB) core 1 App A A TLB A Translation core 2 core 2 Huge Page App Request core 3 core 3 App (1 GB) core 4 core 4 B B core 5 core 5 App App core 6 core 6 4 KB Page C core 7 core 7 Fetch Page Table C 4 KB Page core N core N from Memory If translation not in cache fetch page table from memory and populate TLB https://networkbuilders.intel.com/docs/kilo-a-path-to-line-rate-capable-nfv-deployments-with-intel-architecture-and-the-openstack-kilo-release.pdf

Cloud Native SW w/ Enhance Platform Awareness (EPA) (2/3) Logical Physical Manifestation • Multiple networks and high throughput I/O for DP • Multus CNI plugin and SR-IOV CNI plugin (enables VFs + DPDK user space drivers) https://builders.intel.com/docs/networkbuilders/enabling_new_features_in_kubernetes_for_NFV.pdf

Cloud Native SW w/ Enhance Platform Awareness (EPA) (3/3) Native – Bare metal processes, no containers, no orchestration • K8s – Docker containers orchestrated by K8s with EPA knobs ON / OFF • Test User Space CPU Huge Page Pkts/sec* (w/ noise) Driver Pinning Native Yes Yes Yes 1,550K (1,100K) K8s Yes Yes Yes 1450K (1.150K) K8s No Yes Yes 750K (650K) K8s Yes No Yes 1450K (400K) K8s Yes Yes No 1200K (1100K) *50K Granularity, 1 CPU Core Cloud Native SW w/ EPA achieves performance similar to bare-metal processes • Supporting additional features like AF-XDP, DDP (Device Data Personalization) • Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more information go to http://www.intel.com/performance .

Build Deployment in Aether : Enterprise Edge-as-a-Service Cloud GMA Internet gaming Core Control Plane Monitoring Automation Workflow Service Control HSS HSS DB Mgmt & MME PCRF SMF / SPGW-C Aether Mgmt* Core USER Plane 5G UPF / SPGW-U Aether Mgmt* Enterprise Edge Site Private/Public Central Cloud or Central Office Intel SGX based secure container Operational Cloud Native, Scalable & Distributed Gateways, with central control in private/public clouds (Google / Azure) • Multiple Aether edges deployed e.g. AT&T, NTT, Telefonica, Argela, Ciena, Intel, ONF – More to come • To be deployed as part of DARPA “Verifiable Closed Loop Control Network” with Stanford, Princeton and Cornell • ONF acts as “operator” for Day-0, Day-1, Day-2 (Deployment, reliability, support) - Benefit platform maturity • Deploy and evaluate benefits of edge applications or capabilities, e.g. Cloud Gaming, GMA, etc • * : Aether SW developed by ONF

Edge Service - Cloud Gaming Edge based deployment models Today: Interactive Frame Streaming Model s e t a t s e Game instances Game instances m Game instances Game instances Game instances a Game instances G E2E latency about ~166ms @1080p Edge Edge … Key Challenges: (Rendering) (Rendering) • Reduce E2E Latency (as close to client gaming <70ms) Low latency High bandwidth … … Client Client Client Client • Improve Quality (4K or higher without aliasing / encoding artifacts) • Provide constant throughput (60 or higher FPS) Source: Selvakumar Panneer, Intel Labs