MetroHaul WP 2 Year 2 Review Albert Rafel

Metro-Haul WP 2: Year 2 Review Albert Rafel (BT) Brussels, 2 October 2019 METRO-HAUL: METRO High bandwidth, 5 G Application-aware optical network, with edge storage, compute and low Latency http: //metro-haul. eu H 2020 -ICT-2016 -2 Metro-Haul Grant No. 761727

WP 2 – Use Cases, Service Requirements, and Network Architecture Definition v WP 2 Objectives q to define Service Use Cases and requirements of the metro network architecture that can support them q to define the functional architecture of the metro network that can be used by an orchestrator to control the infrastructure in order to provide the defined Service Use Cases q to design the Metro network architecture and evaluate its performance v WP 2 Work Organisation – Tasks q Task 2. 1 Definition of generic use cases and service requirements [June ‘ 17 – Jan ‘ 18] Leader: TIM ü Milestone: MS 3 “Collection of selected proposals on use cases and services” [finished Sept ‘ 17] ü Deliverable: D 2. 1 “Definition of Use Cases, Service Requirements and KPIs” [submitted Feb ‘ 18] q Task 2. 2 Functional network architecture definition [June ‘ 17 – Sept ‘ 18] Leader: University of Bristol ü Milestone: MS 14 “Architecture Functional Description” [finished Oct ‘ 18] ü Deliverable: D 2. 2 “Functional Architecture Specifications and Functional Architecture” [submitted Oct ‘ 18] q Task 2. 3 Network architecture design & evaluation [Jan ‘ 18 – Nov ‘ 19] Leader: Infinera (formerly Coriant) Portugal ü Deliverable: D 2. 3 “Network Architecture Definition Design Methods and Performance Evaluation” [submitted May ‘ 19] ü Deliverable: D 2. 4 “Techno-Economic Analysis & Network Architecture Refinement” [due Nov ‘ 19] 14/12/2021 Metro-Haul Year 2 Review 2

WP 2 Approach – 5 G Service Classes Distributing intelligence in the network • a single physical network that is dynamically programmed to deliver not only u. RLLC, but also m. MTC and e. MBB services at the same time. • The network uses the efficiencies and computing/storage scalability of data centres – the cloud – at multiple levels: near the edge, at the aggregation level, near the core. • Software-defined networking (SDN) and network function virtualization (NFV) provision a dynamically-configurable, service-oriented architecture. Source: http: //5 gsmarts. com/5 g-tech/5 g-latency-1 -millisecond/ Source: ETSI presentation - Adrian Scrase - URLLC 2017, London, UK, 14 November 2017 14/12/2021 Metro-Haul Year 2 Review Central Offices re-architected as Data Centres 3

Metro-Haul Architecture Control / Orchestration Fixed & Wireless Access Monitoring DATA CENTRE AMEN MCEN BBU EPC BBU DCN Metro Node PBytes? Gbit/s? FLOPS? Photonic Core DC DC DC TBytes? Tbit/s? FLOPS? Metro Node 10 -20 km 0. 1 -0. 2 ms Metro-Haul scope 50 -200 km 0. 5 -2 ms AMEN: Access Metro Edge Node MCEN: Metro Core Edge Node 100 -1000 km 1 -10 ms 14/12/2021 Metro-Haul Year 2 Review 4

Analysed Service Use Cases Vertical Use Case Co. S Content Delivery Network e. MBB Live TV Distribution e. MBB + URLLC 6 Do. F Virtual Reality e. MBB + URLLC Crowdsourced Video Broadcast URLLC Service Robotics e. MBB + URLLC Enterprise Access with NG Eth e. MBB + URLLC Utilities Utility Metering m. Io. T Automotive Intelligent Transport System & Autonomous Driving e. MBB + URLLC Industry 4. 0 Smart Factories e. MBB + URLLC + m. Io. T Media & Entertainment Cloud Services Public Safety & Environment Real Time Low Latency Object Tracking & Security Operator orientated 14/12/2021 Secure SDN Control – Video Distribution URLLC BB + URLLC 5 G Fronthaul Architectures Metro-Haul Year 2 Review 5

D 2. 2 Functional Architecture Specifications & Functional Definition v D 2. 2 binds together the work done in D 2. 1/WP 2, D 3. 1/WP 3 and D 4. 1/WP 4 in terms of SDN control and monitoring, AMEN/MCEN nodes, and optical layer, in scope of the objectives of METRO-HAUL and the Service Use Cases v Metro-Haul uses and EXTENDS existing SDN/NFV platforms (Standards) to achieve the overall objectives q. New models, interfaces and agents (e. g. disaggregated optical nodes) q. Advanced Planning & Optimisation Tools q. Monitoring and Data Analytics Subsystem v D 2. 2 described the functional Workflow and the Performance Analysis for the two field demos and one portable demo as described in D 2. 1/WP 2 & D 5. 1/WP 5 q. Crowd-Source Video Streaming q. Real-Time Low-Latency Object Tracking & Security q. It includes other vertical services described in D 2. 1/WP 2 14/12/2021 Metro-Haul Year 2 Review 6

Deliverable D 2. 2 on Latency v D 2. 2 includes an analysis and classification of the latency requirements for the different services defined in D 2. 1/WP 2 v It also elaborates on a Latency Model and applies it to the Metro-Haul Architecture q It includes a detailed analysis of the latency for the different Latency Classes over different 5 G Transport Architecture Split Options and their mapping onto the Metro-Haul Architecture 14/12/2021 Metro-Haul Year 2 Review 7

Deliverable D 2. 2 on Energy Consumption v D 2. 2 includes an Energy Consumption Analysis applied to the Content Delivery Network (CDN) use case v The model is used to estimate the energy consumed for transporting and storing the service traffic 14/12/2021 Metro-Haul Year 2 Review 8

T 2. 3 Network architecture design & evaluation v Antonio Eira (Infinera) organised work in T 2. 3 through main 3 activities: Project KPIs, Support Tasks, and Working Streams q D 2. 3 Network Architecture Definition, Design Methods and Performance Evaluation q D 2. 4 Techno-Economic Analysis & Network Architecture Refinement Figure 5 / D 2. 3 Deliverable D 2. 3 describes the tools and methodologies to evaluate the Metro-Haul architecture based on the KPIs 14/12/2021 Metro-Haul Year 2 Review 9

Metro-Haul KPIs and T 2. 3/WP 2 M-H KPIs MH 1 Optical Pt. P connection set-up time MH 2 Metro-Haul E 2 E Pt. P connection set-up time MH 3 Set-up time of network service slice across M-H MH 4 Capacity of Metro-Haul controller MH 5 Fault/degradation detection time MH 6 Capacity of Metro-Haul infrastructure MH 7 New optical components/systems MH 8 Cap. Ex reduction MH 9 Energy consumption • Project KPIs were defined by Consortium • Evaluation spread across WPs according relevance • MH 6, MH 8, and MH 9 were partly assigned to T 2. 3/WP 2 M-H KPIs 5 G-PPP Performance KPIs 5 G PPP Performance KPIs P 1 P 3 P 4 P 5 P 6 MH 1 X X X N/A MH 2 X X X N/A MH 3 X MH 4 X P 1 Providing 1000 times higher wireless area capacity and more varied service capabilities compared to 2010. P 2 P 3 Saving up to 90% of energy per service provided. Reducing the average service creation time cycle from 90 hours to 90 minutes. MH 6 X MH 7 X P 4 Creating a secure, reliable and dependable Internet with a “zero perceived” downtime for services provision. MH 8 X P 5 P 6 Facilitating very dense deployments of wireless communication links to connect over 7 trillion wireless devices serving over 7 billion people. Enabling advanced user-controlled privacy. 14/12/2021 Metro-Haul Year 2 Review P 2 X X MH 5 MH 9 X X N/A N/A 10

Geo-Types, Reference Topologies and Traffic Characterisation D 2. 3 defines a reference topology to be used for the validation of the Metro-Haul KPIs Topology as the Consolidation of current TIM, Telefónica and BT Metro Networks Two types of traffic generating methodologies have been developed - Macro Traffic: maximum traffic generated during busy hour - Micro Traffic: dynamic traffic being generated within the busy hour Traffic has been characterised and different methodologies to generate the traffic have been developed. The traffic matrices resulting from these methodologies will be the Traffic Inputs for the Techno-Economic Studies and evaluation of the M-H KPIs 14/12/2021 Metro-Haul Year 2 Review 11

Evaluation Tools v Net 2 Plan as a Tool for the network Techno-economic studies and evaluation q The existing Open Source Planning tool Net 2 Plan was chosen for the use in WP 2 q The necessary tool new interfaces were developed to be used by Metro-Haul v Node Cost and Energy Models q Models to evaluate the cost and power consumption of the optical, packet-switched and data centre building blocks to be used to design Metro-Haul’s AMEN and MCENs 14/12/2021 Metro-Haul Year 2 Review 12

Candidate Architectures & Evaluation Methods v Data-Centre Nodes at AMEN/MCEN q Six different architectures have been defined that will be assessed q ADVA commercial equipment has been adapted and enhanced to be used in the DC Node as computing and switching resource v Disaggregated Optical node architectures for Metro-Aggregation and for Metro-Core Nodes q Physical Layer Components developed within WP 3 q Optical Disaggregation Models q Edge Computing v Autonomous Networking & E 2 E Orchestration Metro-Haul topology and full network topology 14/12/2021 Metro-Haul Year 2 Review 13

D 2. 4 Techno-economic Analysis and Network Architecture Refinement v D 2. 4 due November 2019, will show the “Make or Break” of WP 2 1. Frameworks for Techno-Economic Architecture Evaluation q Traffic Characterization and Service Use-Cases (UPC) q Planning Tools and Dissemination (UPCT) q Cost and Energy Model (UC 3 M) 2. Architecture and Evaluation Methods Refinement q q Techno-Economic Analysis of DC Node Architectures at Central Offices (TID) Physical Network Architectures for Optical Nodes (INFN) Economic Impact of Optical Disaggregation Models (TIM) Techno-Economic Impact of Edge Computing (Poli. Mi) 3. Techno-economic evaluations and KPI analysis (INFN) 4. Conclusion 14/12/2021 Metro-Haul Year 2 Review 14

Techno-economic analysis. Methodology 14/12/2021 Metro-Haul Year 2 Review 15

Methodology for techno-ecnonomic analysis Scenario Definition Reference Topology Traffic Model Software framework NIW Excel Template NFV over IP over WDM Net 2 Plan Library Reusable modules / multiple partners Algorithms Aut. reports Benchmark tests scripts • NIW library developed within MH (open-source) • Based on abstract model of an IP over WDM network with IT resources in the nodes. • Simplifies development of Net 2 Plan algorithms, automatic reports etc. for these networks. • Simplifies import/export from defined Excel-based template. • Publicly available: shipped with Net 2 Plan (www. net 2 plan. com) 12/14/2021 Metro-Haul Year 2 Review 16

Methodology for techno-economic analysis. Workflow example IP BOM Reference Topology & Traffic Capacity Planning alg. DC BOM WDM BOM Cost model Energy model Multi-hour tests Multi-period tests • Three reference topologies (Telecom Italia) • Diameter <200 km, [52, 102, 159] nodes • Anonymized in node position • With traffic information in agreement with Deliverable 2. 3 • Background traffic: P 2 P, regular, cacheable (video) • Traffic from verticals: VNF & traffic requisites from key vertical services considered (D 2. 3). 12/14/202 1 Metro-Haul Year 2 Review 17

Methodology for techno-economic analysis. Workflow example Cost model IP BOM Reference Topology & Traffic Capacity Planning alg. DC BOM WDM BOM RUN 12/14/202 1 Energy model Multi-hour tests Multi-period tests • JOINT capacity planning of IT & network resources • Different algs from different partners for different problem variants / approaches • ML algorithms in resource allocation in dynamic traffic environments • Code reuse: Algorithms can leverage on proven routines or other algorithms Metro-Haul Year 2 Review 18

Methodology for techno-economic analysis. Workflow example Cost model IP BOM Reference Topology & Traffic Capacity Planning alg. DC BOM WDM BOM Energy model Multi-hour tests Multi-period tests Automatic scripts creating Bill-of-Materials reports • IT: BOMs for different DC architectures • IP & WDM: BOMs for different capabilities of optical equipment 12/14/202 1 Metro-Haul Year 2 Review 19

Methodology for techno-economic analysis. Workflow example IP BOM Reference Topology & Traffic Capacity Planning alg. DC BOM WDM BOM Datacenter Components Extra. Large Storage SSD Small SSD Large NAS Small NAS Medium NAS Large NAS Extra. Large Examples: Samsung SSD, i. XSystems True. NAS X 10 2 U, HPE Nimble Storage HF 40/60, Dell EMC Unity 300 4 TB 8 TB 20 TB 60 TB 120 TB 400 TB Small Medium Large Other specialized hardware BRAS HW Firewall Load Balancer 12/14/202 1 Energy model Multi-hour tests Multi-period tests Description (text) Examples: HP Proliant DL 360, Huawei 1288 -v 5 or 2288 HV 5, Dell R 740 or R 940 Intel Xeon Gold 6134 with 8 cores, 64 GB RAM, 600 GB HDD Intel Xeon Gold 6140 with 16 cores, 128 GB RAM, 1. 2 TB HDD Intel Xeon Platinum 8160 with 2 CPUs x 24 cores, 128 GB RAM, 3. 6 TB HDD Intel Xeon Platinum 8160 with 4 CPUs x 24 cores, 192 GB RAM, 3. 6 TB HDD Compute Nodes Cost model eg Fortinet Forti. Gate 3000 D eg HPE Open. Call Load Balancer, Radware Alteon. F 5 Big. IP, etc MH Cost model & energy consumption model • A cost model is being produced, updating and expanding efforts in previous Eus • Cost model includes energy consumption figures • Automatic reports are being developped to produce cost & energy figures from the BOMs & networks. Metro-Haul Year 2 Review 20

Methodology for techno-economic analysis. Workflow example IP BOM Reference Topology & Traffic Capacity Planning alg. DC BOM WDM BOM Cost model Energy model Multi-hour tests Multi-period tests Different tests are being conducted. Scripts for bulk definition of the multiple tests can be also shared. • Multi-hour: Exploit knowledge of well known traffic profile variation along day. • Multi-period: Studies considering traffic growth. Future-proof network evolution. Relevant results already obtained, described later in this talk 12/14/202 1 Metro-Haul Year 2 Review 21

Static traffic matrices Traffic model proposal based on three main categories A. Residential and Business «mass market» . Three types: 1. (10%) Peer-to-peer (audio and video call and other services between user terminals) – Symmetrical (Upstream 100% of Downstream) ü User ↔ MCEN BB (5 G Core handling IP packets) ↔ User 2. (30%) No cached traffic from remote DC (in Metro Core BB sites or outside) - no caching in metro network assumed – Asymmetrical (Upstream 30% of Downstream) ü User↔MCEN BB (where application DCs are, or going to the core) 3. (60%) Cached video traffic on remote or local DC Hugely asymmetrical (Upstream 10% of Downstream) ü User↔AMEN Cache ü User↔Closest MCEN Cache ü User↔MCEN BB (Cache / go to backbone) Traffic intensities: short, medium and long term traffic (2019, 2022, 2025), reflecting also expected 5 G penetration 14/12/2021 Metro-Haul Year 2 Review 22

Daily Traffic Profile For all the traffic types: • 80% residential nature • 20% business nature Residential and business have different daily patterns Note that residential traffic flows peak at night, and business traffic flows peak in the afternoon 14/12/2021 Metro-Haul Year 2 Review 23

Static vs Dynamic Control & Data Planes • Static case: • Traffic flows of residential and business traffic separated • No reconfiguration possible: • IP flows are statically assigned resources for peak load. • Lightpaths are statically placed, never changed, but shared between residential and business traffic • Dynamic case. Metro. Haul agile control and data plane permits: • Make the most of Physical NFs VNFs • Redimension the IT resources assigned in the DCs to each VNF • IT resources used in the morning for NFs of business traffic, can be reused in the evening for NFs handling residential traffic. • Lightpaths are set up and down dynamically on demand to adapt to traffic variations • Transponders INSIDE a node are reusable: A transponder in node A can be used in the morning for a lightpath A↔B, and later in a lightpath A↔C. 14/12/2021 Metro-Haul Year 2 Review 24

Technoeconomics Outcomes in Topology Dense Having a dynamic & agile control & data plane helps to reduce CAPEX? TOPOLOGY 1 Static Dynamic TOPOLOGY 2 Static Dynamic Tp 1 0 16 Tp 1 0 71 Server (T 1) 20. 5 17 Server (T 1) 29. 6 25 TP 2 TPOther 466 756 386 630 TP 2 TPOther 5500 468 4588 370 Cache (T 2) 34. 7 28. 9 Cache (T 2) 50. 1 41. 6 UPF (All) 65. 7 54. 6 UPF (All) 94. 9 78. 8 20 -27% reduction in # transponders needed, both in filterless and full ROADM cases (Transponders - TP 1 – 10 G, TP 2 25 G, TPOther 100 G) 17%-20% reduction in IT resources CAPEX estimated as the amount of Tbps of traffic to be processed by the VNFs thanks to their dynamic resizing

Transponders used in MCEN and AMEN Nodes (Static & Dynamic) Dynamic case also helps in energy consumption OPEX (# active lightpaths used) 14/12/2021 Metro-Haul Year 2 Review 26

Traffic per VNF Type in MCEN and AMEN Nodes (Static & Dynamic). Dense Dynamic case also helps in energy consumption OPEX (reserved (even if idle) capacity in IT resources) 14/12/2021 Metro-Haul Year 2 Review 27

SUMMARY 14/12/2021 Metro-Haul Year 2 Review 28

Summary v Very successful, though challenging, first 2 years of WP 2 activities q M 2. 1 & D 2. 1 very important input to all WPs and activities in Metro-Haul q M 2. 2 & D 2. 2 brought together all the initial work done in WP 2, WP 3, WP 4, and WP 5 ü Huge effort from many partners from within and outside WP 2 ü Some partners without planned effort in WP 2 made significant and important contributions q T 2. 3 is a very complex and critical activity cleverly and efficiently led by Infinera ü D 2. 3 laid the Foundations (Tools & Methodology) of how to evaluate the Metro-Haul architecture & different solution options q D 2. 4 will show some of the Key Results of Metro-Haul Assessment & Evaluation v Use of resources q Activities have exceeded what had been planned in the proposal (e. g. development of Traffic Generating Tools) q Higher complexity that anticipated (e. g. analysing the Use Cases) q Some partners have made much larger contributions than anticipated (e. g. UPC/UPCT) v NEXT STEPS q Deliver D 2. 4 q WP 2 planned to stop activities by end of November 2019 14/12/2021 Metro-Haul Year 2 Review 29

Conclusions & Next Steps v Very important inputs to all WPs and activities in Metro-Haul v Bringing together all the initial work done in WP 2, WP 3, WP 4, and WP 5 v Edge-to-edge fully disaggregated Metro-Haul architecture finalised assimilating physical and control plane aspects v Laid the Foundations (Tools & Methodology) to enable evaluation of the Metro-Haul architecture & different solution options v. Techno-economic analysis with well-defined use-cases assessing the physical layer solutions already showing benefits v NEXT STEPS q Deliver D 2. 4 q WP 2 planned to stop activities by end of November 2019 14/12/2021 Metro-Haul Year 2 Review 30

Questions or Clarifications? THANK YOU http: //metro-haul. eu 14/12/2021 Metro-Haul Year 2 Review 31