Carrier SDN for Multilayer Control

Similar documents
ABNO: a feasible SDN approach for multivendor. optical networks. A. Aguado, V. López, J. Marhuenda, O. González de Dios and J.P. Fernández-Palacios

Juan Fernandez-Palacios Telefónica I+D. Carrier SDN Control plane and SDN convergence in E2E networks

Why do operators need multi-layer coordination?

Packet-Optical SDN Field Trial for Multi-Layer Network Optimization. Jim Theodoras May 2016

Coriant Transcend Symphony Solution

Latest work in the IETF for SDN and NFV

Software defined networking

Functional validation of the cooperation between Virtual Network Topology Manager and Path Computation Element

End-to-end service provisioning across MPLS and IP/WDM domains

NETFUSION DISCOVERY SYSTEM DESCRIPTION

Open Source Netphony Suite: Enabling Multi-layer Network Programmability

Cisco WAN Automation Engine (WAE) Network Programmability with Segment Routing

SDN Evolution of networks. Raul Caldeira

Infinera Intelligent Transport Network A new architecture for the Terabit Era

OPEN CONTRAIL ARCHITECTURE GEORGIA TECH SDN EVENT

SDN for Multi-Layer IP & Optical Networks

Automated Control and Orchestration within the Juniper Networks Mobile Cloud Architecture. White Paper

BROCADE CLOUD-OPTIMIZED NETWORKING: THE BLUEPRINT FOR THE SOFTWARE-DEFINED NETWORK

SDN Commercial Deployments: Emerging Business Cases

Network Planning, Control and Management Perspectives on Dynamic Networking

Getting started with O3 Project Achievement ~ Innovating Network Business through SDN WAN Technologies~

Open, Hierarchical SDN Control Systems: Ecosystem Progress

Innovative and Open Network Architectures for research and education networks and automated operation through SDN

The Role of the Path Computation El ement Centralized Controller in SDN & NFV

Programmable BitPipe. Andreas Gladisch VP Convergent Networks and Infrastructure, Telekom Innovation Labs

O3 Project Network Business Innovation by SDN WAN Technologies

Routing Applications State of the Art and Disruptions

SDN Solution for Service Provider Access Network Dennis Pai, Product Manager Ahmed Abeer, Technical Marketing Engineer BRKSPG-2064

MULTILAYER REQUIREMENTS FOR NBI OF IP/MPLS DOMAIN CONTROLLERS

Decoding MPLS-TP and the deployment possibilities

Transformation through Innovation

NETWORK DEPLOYMENT WITH SEGMENT ROUTING (SPRING)

MPLS network built on ROADM based DWDM system using GMPLS signaling

Current Trends in IP/Optical Transport Integration

Open and Disaggregated Transport SDN

NSP Network Services Platform Release Deployment Overview. 3HE AAAA-TQZZA Issue 1 March 2018

SDN Controllers in the WAN: protocols and applications

Layer 1, 2 and 3 Integration

Digital Transformation

Estratégias de integração entre as camadas IP e Óptica The Path to Efficient Networks

DECODING SOFTWARE DEFINED NETWORKS

SDN Software-Defined Networking

A Possible New Dawn for the Future GÉANT Network Architecture

China Unicom SDN Practice in WAN. Lv Chengjin/Ma Jichun, China Unicom

GÉANT Network Evolution

Future of PCE. Dhruv Dhody, Sr. System Architect, Huawei Technologies

Cisco PONC Walid Wakim Principal Engineer. March 10, 2015

Simplified service creation and delivery. Branch. SOHO Data Center. Control Center / NOC Packet Muse Service & Network Applications

Optical network virtualization. Optical network virtualization. Transport SDN for cloud-centric networking

Pradeep Kathail Chief Software Architect Network Operating Systems Technology Group, Cisco Systems Inc.

EXPLORER SDN PATH PROVISIONING

APNIC elearning: SDN Fundamentals

NTT Communications' Perspective on Next GEN Optical Transport Network

Unlock the Benefits of Transport SDN OIF Transport SDN API Interop Demo

THE MPLS JOURNEY FROM CONNECTIVITY TO FULL SERVICE NETWORKS. Sangeeta Anand Vice President Product Management Cisco Systems.

Overview on FP7 Projects SPARC and UNIFY

Benefits of SDN Modeling and Analytics tool for complex Service Provider Network

Technologies for the future of Network Insight and Automation

Transport SDN: The What, How and the Future!

Multilayer Design. Grooming Layer + Optical Layer

The Next Opportunity in the Data Centre

Introduction to IEEE 802.1Qca Path Control and Reservation

The Emerging Optical Control Plane

Making Inroads to Infrastructure SDN with Cisco Quantum WAVE. Sonny Franslay Consulting SE, WAN Orchestration

SDN Controllers in the Broadcast TV Environment

Ericsson ip transport nms

Inter-SDN in Seamless MPLS for Mobile Backhaul

Technical White Paper on SDN WAN

OpenFlow: What s it Good for?

Core Networks Evolution

AGILE OPTICAL NETWORK (AON) CORE NETWORK USE CASES

Modernização da Rede IP (Docsis 3.1), Fotônica e SDN

Nathan Tracy OIF Technical Committee Chair ECOC 2016 Market Focus September 20, 2016

Karl Gass OIF PLL Vice Chair - Optical Track NGON 2016 July 1, 2016

SD-WAN 101. November 3 rd 2016 Rob McBride Marketing

ODTN An Open Controller for the Disaggregated Optical Network

FP7 EU-JP Project : STRAUSS. Ken ichi Kitayama (JP Project Coordinator) Osaka University

Trends in Optical Disaggregation. Presented by :

Versa Software-Defined Solutions for Service Providers

ONF SDN Architecture and Standards for Transport Networks

Optical + Ethernet: Converging the Transport Network. An Overview

MASERGY S MANAGED SD-WAN

SDN TO BE OR NOT TO BE. Uwe Richter SE Director Russia/CIS, East and South East Europe

Data Center Revolution Impact on Ethernet and MPLS

The CORD reference architecture addresses the needs of various communications access networks with a wide array of use cases including:

WELCOME. Chicago Juniper Users Group SEPT 18TH, 2013

Taxonomy of SDN. Vara Varavithya 17 January 2018

Service-centric transport infrastructure

The TeraStream Approach - A Means for the provision of Cost Efficiency and Service Integration

Transport SDN and Use Cases in Korea

Emerging Technologies in Optical Transport Networks

Coriant Groove G30 Network Disaggregation Platform

The Consumable. Preparing business networks for the cloud with Software Defined Networking

Release Notes for NorthStar Controller

Adventures in Multi-Layer, Multi- Vendor Network Control. Wes Doonan Control Plane R&D July 2010

History of SDN Control at AT&T

Hard Slicing: Elastic OTN and Wavelength Slicing

Transport SDN at OIF Assuring a Seamless Evolution to Interoperable Transport Networks of the Future

Introduction to Segment Routing

Unified MPLS for Multiple Applications Transport Profile

Transcription:

Carrier SDN for Multilayer Control Savings and Services Víctor López Technology Specialist, I+D Chris Liou Vice President, Network Strategy Dirk van den Borne Solution Architect, Packet-Optical Integration 1

Rationale Changes in Traditional Networks Traditional carriers networks operation is very complex and is neither readily adaptable nor programmable to traffic changes. Multiple manual configuration actions are needed in metro and core network nodes. Network solutions from different vendors typically use vendor-specific Network Management System (NMS) implementations. Software Defined Networking (SDN) and network programmability offer the ability to direct application service requests towards the IP/MPLS and optical network, but the proposed SDN controllers in the market are based on monolithic software, which are not adapted to current heterogeneous network environments. 3 Rationale Why do operators need multi-layer coordination? 1 Automated multilayer coordination Network resources optimization (CAPEX reduction) Operation simplification (OPEX reduction) 2 Multilayer Restoration Enables the reduction of the number of ports for protection. Improves network availability and the MTTR. 3 Rapid Service Delivery Multi-layer provisioning enables faster deployment Open platform with programmability that spurs app innovation 4 2

FEATURE FEATURE OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATUR FEATUR E E OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATURE FEATURE OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATURE FEATURE OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATUR FEATUR E E OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATUR FEATUR E E OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATURE FEATURE OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE FEATUR FEATUR E E OPERATING SYSTEM SPECIALIZED PACKET FORWARDING HARDWARE 05/05/2014 Carrier SDN architecture Software Defined Networking SDN Network equipment as Black boxes Open interfaces (OpenFlow) for instructing the boxes what to do Boxes with autonomous behaviour SDN Decisions are taken out of the box Adapting OSS to manage black boxes SDN Simpler OSS to manage the SDN controller 6 3

02 Which architecture fits for Carrier SDN? SDN CONTROLLER Big black box controlling the network Centralize functionalities to enable automation Define simple standard interfaces 7 ABNO architecture To design and prototype a new control architecture to enable automated and simplified layer network service provisioning through different network segments (metro, core, data center ) and technologies (IP/MPLS, optical, OpenFlow ) Network configuration points minimization by transferring multidomain and multilayer provisioning functionalities from Network Management Systems (NMS) to distributed signaling mechanisms (control plane). Unified network configuration and orchestration mechanisms From proprietary CLI to standard configuration interfaces (e.g NetConf, OpenFlow ). SDN architecture enabling end to end network orchestration according to service and network optimization criteria. 8 4

Telefónica Multi-Layer SDN Architecture Proof of Concept Network as a Service (NaaS) Application: Dynamic MPLS tunnel service creation in multilayer, multi-vendor environment. ABNO Multi-Layer SDN Control Layer MX-240-2 Multi-layer PCE-based controller: Point and click IP/MPLS services w/automatic router & transport layer provisioned automatically MX-240-1 10G 10G OTS DTN1 100G 10G 10G OTS DTN2 100G 100G MX-240-3 10G 10G OTS DTN3 Multiple south-bound protocols REST/JSON OpenFlow Netconf/YANG PCEP BGP-LS 9 Network as a Service in Multilayer Environments Each user sees logically partitioned view MPLS paths requested through single user interface Automated multilayer provisioning is done at the IP/MPLS and OTN layers by ABNO controller Live Demos Performed: Telefonica Labs Open Network Summit 2014 OFC 2014 Infinera s booth Video of demo http://www.infinera.com/go/sdn Infrastructure View Per-User View Juniper MX-240 Infinera DTN 10 5

ABNO Multi-layer SDN Control Layer SDN Control Layer based on IETF building blocks Applications (Internet, CDN, cloud ) ALTO ABNO controller OAM Handler TED VNTM PCE Open SDN Control Layer Provisioning Manager OPENFLOW REST BGP/LS NETCONF PCEP CLI OpenFLow Data Center OpenFlow MAN Domain IP/MPLS core GMPLS Optical Domains OpenFlow Optical Domain MPLS MAN 12 6

ABNO building blocks ABNO Controller Main component of the architecture Responsible of orchestrating Request from the NMS/OSS and selects the appropriate workflow to follow in order to satisfy each request. Path Computation Element Path computation across the network graph. Coordination between multiple PCEs for multi-domain (for example, inter-as) or multi-layer networks. Instantiation capabilities. OSS/NMS / Applica on Service Orchestrator Policy ABNO Controller Agent OAM Handler ALTO VNTM Server I2RS Client PCE Topology Module Provisioning Manager Client Network Layer Server Network Layer 13 ABNO building blocks Virtual Network Topology Manager (VNTM) Maintaining the topology of the upper layer by connections in the lower layer. Simplifies the upper-layer routing and traffic engineering decisions. Topology Module. Retrieve and provide network topology information, both per-layer topologies as well as inter-layer topology. Provisioning Manager. In charge of configuring the network elements so the LSP can be established. There are several protocols that allow the configuration of specific network resources such as Openflow, Netconf, CLI and PCEP. OSS/NMS / Applica on Service Orchestrator Policy ABNO Controller Agent OAM Handler ALTO VNTM Server I2RS Client PCE Topology Module Provisioning Manager Client Network Layer Server Network Layer 14 7

CARRIER SDN FOR MULTI-LAYER CONTROL APPLICATION-DRIVEN TRAFFIC ENGINEERING CARRIER SDN THE MOTIVATION FOR SDN IN WIDE AREA NETWORKS Benefits of SDN in the WAN End-to-End path computation : Optimize utilization of network resources Predictable network behavior Application-aware traffic engineering Multi-layer convergence: Control plane integration between packet and transport layers for automated multi-layer service provisioning of end-to-end paths. Optimize utilization of network resources for protection through 2 nd order restoration, prevent network flapping SDN controller Network CHALLENGE How to combine the resilience of distributed networking with the predictability of centralized control? 8

SDN IN WIDE AREA NETWORKS CENTRALIZE WHAT YOU CAN, DISTRIBUTE WHAT YOU MUST Centralized TE controller in a packet-transport WAN network: Performs TE path Discovery Provisioning Monitoring Modification Provides centralized processing for TE database Highly scalable architecture for very granular deployment of traffic engineering. Operators get full control of routing policy. Network operates much as today: Network elements retain signaling function. Local RIB/FIB based forwarding. Use of established and widely deployed routing protocols. WAN SDN CONTROLLER: NORTHSTAR APPLICATION-DRIVEN TRAFFIC ENGINEERING SDN controller use cases Bin packing or network defragmentation. Differentiated traffic engineered Paths, e.g. guaranteed lowest latency or lowest cost. Bandwidth scheduling / calendaring. Predictable traffic steering / path diversity. Adaptive TE control loops Automated re-signaling before / after maintenance windows. Offline network planning STANDARDIZED PROTOCOLS AND OPEN INTERFACES ANALYZE OPTIMIZE VIRTUALIZE Topology Discovery Routing PCEP - LSP discovery IGP-TE, BGP-LS - TED discovery LSP statistics - Analytics Path Computation Optimization algorithms PCEP - Create traffic engineered LSP Path Installation PCEP Netconf / YANG Netconf / YANG Configuration (may include: BGP, OpenFlow, I2RS) 9

WAN SDN CONTROLLER: NORTHSTAR APPLICATION-DRIVEN TRAFFIC ENGINEERING ABNO controller architecture OSS / NMS / Application Service Coordinator STANDARDIZED PROTOCOLS AND OPEN INTERFACES Policy Agent ALTO Server Databases TED LSP-DB VNTM ABNO Controller PCE Provisioning Manager Client Network Layer Server Network Layers I2RS Client OAM Handler ANALYZE OPTIMIZE VIRTUALIZE Topology Discovery Routing PCEP - LSP discovery IGP-TE, BGP-LS - TED discovery LSP statistics - Analytics Path Computation Optimization algorithms PCEP - Create traffic engineered LSP Path Installation PCEP Netconf / YANG Netconf / YANG Configuration (may include: BGP, OpenFlow, I2RS) WAN SDN CONTROLLER: NORTHSTAR APPLICATION-DRIVEN TRAFFIC ENGINEERING ABNO controller architecture OSS / NMS / Application Service Coordinator STANDARDIZED PROTOCOLS AND OPEN INTERFACES Policy Agent ALTO Server Databases TED LSP-DB VNTM ABNO Controller PCE Provisioning Manager Client Network Layer Server Network Layers I2RS Client OAM Handler ANALYZE OPTIMIZE VIRTUALIZE Topology Discovery Routing PCEP - LSP discovery IGP-TE, BGP-LS - TED discovery LSP statistics - Analytics Path Computation Optimization algorithms PCEP - Create traffic engineered LSP Path Installation PCEP Netconf / YANG Netconf / YANG Configuration (may include: BGP, OpenFlow, I2RS) 10

WAN SDN CONTROLLER: NORTHSTAR APPLICATION-DRIVEN TRAFFIC ENGINEERING ABNO controller architecture OSS / NMS / Application Service Coordinator STANDARDIZED PROTOCOLS AND OPEN INTERFACES Policy Agent ALTO Server Databases TED LSP-DB VNTM ABNO Controller PCE Provisioning Manager Client Network Layer Server Network Layers I2RS Client OAM Handler ANALYZE OPTIMIZE VIRTUALIZE Topology Discovery Routing PCEP - LSP discovery IGP-TE, BGP-LS - TED discovery LSP statistics - Analytics Path Computation Optimization algorithms PCEP - Create traffic engineered LSP Path Installation PCEP Netconf / YANG Netconf / YANG Configuration (may include: BGP, OpenFlow, I2RS) ABSTRACT LINKS TOPOLOGY ABSTRACTION BETWEEN NETWORK LAYERS Traffic engineering reachability information allows the computation of an end-to-end path through a series of connected networks: https://datatracker.ietf.org/doc/draft-farrel-interconnected-te-info-exchange/ For example as abstract link enhancements for the GMPLS Overlay Model to compute an path through packet and transport layer. Key concept is the definition of an abstract-link within the transport domain. Only border nodes connecting packet and transport domain. Contains link metrics, such as cost, latency and bandwidth. Contains information on shared risk link groups (SRLG), i.e. abstract links are categorized in fatesharing groups. No information on full transport topology or optical impairments. R1 R3 o1 o2 o3 o4 o5 Abstract-link = {o1, o6} Establish R1 {o1, o6} R4 o6 R2 R4 11

MULTI-LAYER CONTROL E2E TRAFFIC ENGINEERING IN PACKET AND TRANSPORT LAYERS SDN is a key enabler for packet-transport control plane integration in WAN networks : Packet domain can signal a new paths through the transport domain. End-to-end paths through different network domains and layers. Taking into account fate sharing to guarantee end customer SLAs. R1 R3 1 2 3 4 Network analytics Packet demand pattern R2 R4 Multi-layer service provisioning in Minutes, not Months Topology abstraction o1 o3 o4 o5 E2E path signaling o2 o6 MULTI-LAYER CONTROL E2E TRAFFIC ENGINEERING IN PACKET AND TRANSPORT LAYERS SDN is a key enabler for packet-transport control plane integration in WAN networks : Packet domain can signal a new paths through the transport domain. End-to-end paths through different network domains and layers. Taking into account fate sharing to guarantee end customer SLAs. R1 R3 1 2 3 4 Network analytics Packet demand pattern R2 R4 Multi-layer service provisioning in Minutes, not Months Topology abstraction o1 o3 o4 o5 E2E path signaling o2 o6 12

Role of Intelligent Transport Networking in Carrier SDN 25 Infinera Copyright 2014 Key Trends in Optical Transport IP/MPLS Routers Packet- Optical Transport (P-OTN) Scalability Super-channel Transmission Diverse service rates Programmable transmission Transport Networking Trends 26 Infinera Copyright 2014 13

Key Trends in Optical Transport IP/MPLS Routers Packet- Optical Transport (P-OTN) Scalability Super-channel Transmission Diverse service rates Programmable transmission Transport Networking Trends Convergence Multi-tiered switching OTN, WDM, ROADM, MPLS Mesh topologies 27 Infinera Copyright 2014 Key Trends in Optical Transport IP/MPLS Routers Packet- Optical Transport (P-OTN) Scalability Super-channel Transmission Diverse service rates Programmable transmission Transport Networking Trends Convergence Multi-tiered switching OTN, WDM, ROADM, MPLS Mesh topologies Automation Circuit & Packet service programmability GMPLS (intra-domain) 28 Infinera Copyright 2014 14

Key Trends in Optical Transport IP/MPLS Routers Packet- Optical Transport (P-OTN) Scalability Super-channel Transmission Diverse service rates Programmable transmission Transport Networking Trends Convergence Multi-tiered switching OTN, WDM, ROADM, MPLS Mesh topologies Automation Circuit & Packet service programmability GMPLS (intra-domain) Intelligent transport networks ideal for agile, real-time bandwidth delivery 29 Infinera Copyright 2014 Enabling Open Software Control for Transport Open, modular control layer requires flexible integration model Centralization Monolithic or Universal controller Abstractions key to simplifying programmability Bandwidth programmability Network Element programmability Bandwidth virtualization abstracts physics of optics Nodal & Network-level bandwidth connectivity models Network Element Virtualization enhances SDN integration Decouples control layer from physical NE implementation Simplify integration into SDN framework without additional controller Configuration & Management of service resources Nodal & network-level flows Virtual NE resources (virtual links, virtual ports) & Topology Vendor & implementation agnostic 30 Infinera Copyright 2014 15

Enabling Open Software Control for Transport Open, modular control layer requires flexible integration model Centralization Monolithic or Universal controller Abstractions key to simplifying programmability Bandwidth programmability Network Element programmability Bandwidth virtualization abstracts physics of optics Nodal & Network-level bandwidth connectivity models Network Element Virtualization enhances SDN integration Decouples control layer from physical NE implementation Simplify integration into SDN framework without additional controller Configuration & Management of service resources Nodal & network-level flows Virtual NE resources (virtual links, virtual ports) & Topology Vendor & implementation agnostic 31 Infinera Copyright 2014 Enabling Open Software Control for Transport Open, modular control layer requires flexible integration model Centralization Monolithic or Universal controller Abstractions key to simplifying programmability Bandwidth programmability Network Element programmability Bandwidth virtualization abstracts physics of optics Nodal & Network-level bandwidth connectivity models Network Element Virtualization enhances SDN integration Decouples control layer from physical NE implementation Simplify integration into SDN framework without additional controller Configuration & Management of service resources Nodal & network-level flows Virtual NE resources (virtual links, virtual ports) & Topology Vendor & implementation agnostic 32 Infinera Copyright 2014 16

Enabling Open Software Control for Transport Open, modular control layer requires flexible integration model Centralization Monolithic or Universal controller Abstractions key to simplifying programmability Bandwidth programmability Network Element programmability Bandwidth virtualization abstracts physics of optics Nodal & Network-level bandwidth connectivity models Network Element Virtualization enhances SDN integration Decouples control layer from physical NE implementation Simplify integration into SDN framework without additional controller Configuration & Management of service resources Nodal & network-level flows Virtual NE resources (virtual links, virtual ports) & Topology Vendor & implementation agnostic 33 Infinera Copyright 2014 Bandwidth Virtualization Abstraction of optical transport technologies Optical capacity virtualized into flexible pool of digital bandwidth Bandwidth services completely decoupled from transmission layer 200G available 800G available 650G available 265G available Bandwidth Virtualization Converged Optical Transport 34 Infinera Copyright 2014 17

Open API 05/05/2014 Bandwidth Virtualization Abstraction of optical transport technologies Optical capacity virtualized into flexible pool of digital bandwidth Bandwidth services completely decoupled from transmission layer C A 40GbE bandwidth 100GbE bandwidth 760G available D 610G available B 100G available 165G available Bandwidth Virtualization Converged Optical Transport Bandwidth Virtualization creates pool of shared bandwidth. 35 Infinera Copyright 2014 Open Transport Switch (OTS) Facilitating Transport Virtualization Telefonica ABNO Control Layer REST/JSON OpenFlow OTS OTS Virtual Transport Network OTS Physical Transport Network OTS Open, location transparent SW ABNO integration via OpenFlow & REST Virtual NE abstraction Virtual Node, Virtual Link, Virtual Ports Nodal & Network Flows Multiple provisioning models Direct (nodal) Implicit (network-level, GMPLS) OTS Management & Configuration Discovery & Monitoring Provisioning Mgmt Agent Discovery Agent Physical System Resources Data Agent V i r t u a l i z a t i o n m a p p i n g 36 Infinera Copyright 2014 18

Demonstration Overview and Conclusions Key Takeaways 1 2 CAPEX and OPEX are reducing margins to network operators SDN can help to define abstract interfaces for services thus reducing NMS complexity 3 4 5 SDN can not be a big black box, thus standard building blocks and interfaces are required and SDN should delegate functionalities to control plane Flexible technologies are mandatory in the data plane to allow the network adaptation Horizontal and vertical orchestration allows an automatic network provisioning of network services in layered networks 38 19

20