Hema Kadia

VP - Head of Strategy and Practice for SDN/NFV

Case Study: Orchestration of

Hybrid Networks

Achieving End-To-End Service

Orchestration across Hybrid Networks
Russ Bartels
Director – SDN & Networks Automation/Analytics
 Digital Lifestyle
Our Services:
Windstream
Residential, Small Business, Enterprise and
Wholesale Business Segments • Data Networking
• Core Transport
• Security
• Unified
Communications
SDN Services
Windstream is a U.S. • Managed Services
based FORTUNE 500 • Optical Wave
• SD-WAN
company and a leading
• Carrier Ethernet
provider of advanced • Broadband
network communications • SD-WAN
and technology solutions • Entertainment
• Broadband Element
Commissioning
2
 Who We Are

Key Clientele Portfolio

▪ RPA/Telebots
Exclusive focus Trusted partner to leaders
in the communications ▪ SDN/NFV
on the digital industry, including Fortune ▪ AI/ML & IoT
service provider 500 customers and 95% of
(DSPs) vertical clients have re-engaged ▪ O/BSS and BPO
▪ DevOps &
Microservices

Part of
The Jhaver Group Right-Sized, Flexible Partner
with multi-shore
$1.3 Billion delivery model
35%+ YOY Growth
64 Locations Global Operations
16,500 Employees

3
 1 Reduce service delivery time

2
Revenue & margin growth
Business Drivers
3
for Network Seamless capacity management

Transformation
4 OPEX and CAPEX reduction

5
Enhanced customer experience with self-service

4
 Intent-based, multi-domain service orchestration

Technical Solution
Requirements Flow through automation

Network-based topology database

Interoperability between legacy and SDN-enabled

network appliances

Avoid vendor lock-in

Microservices-based deployment architecture

 Key Considerations:
Architecture Vision for Based Multi-Domain Orchestration
• Cloud infrastructure at core,
edge & premises
O/BSS Stack
a
Services API

Global multi-domain network & • Public & private cloud tight

Closed Loop
service orchestrator Assurance coupling for service chaining
SDN Controller EMS/Controller SD-WAN Controller NFV MANO

ACCESS NETWORK METRO EDGE NETWORK CORE NETWORK • Programmable optical network
PPPoE
International optimizing traffic flows
Peering

IP/MPLS
DSLAM Metro INTERNET
L2 Switch OADM
Packet CORE GIGA ROUTER

POTS
POT Switch
Optical
BNAS
ROUTER
DWDM RING CORE ROUTER • Overarching service
OLT
METRO
L2 Switch
CORE ROUTER
orchestration across physical
SWITCH
OADM
OADM
& virtual network
RADIO
ACCESS ETH CUSTOMER EDGE PROVIDER EDGE
SWITCH
CUST. LAN
 Sample multi-domain, multi-vendor What we did? E.g. Challenges mitigated
ecosystem architecture

Catalog,
Data 1 Northbound • Lack of open APIs for
CPQ & Inventory
Portal
Order
Virtualization
Platform
Management Interfaces (NBI) OSS
Management

Middleware Service Templates • Lack of single source of

REST Microservices REST
Platforms
2 (TOSCA/HEAT) truth e.g. inventory data
Global multi-domain network & service orchestrator
(Service Templates)
• Lack of legacy network
YANG
RESTCONF
Openflow REST CLI REST CLI NETCONF CLI
programmability
SDN Controllers EMS/Controllers
3 Southbound
…. ….
Interfaces (SBI) • IT & network skillset
bridging
Transport Network Access Network DSLAM IP/MPLS Network

7 7
 Switching & routing
4
What we did? E.g. Challenges mitigated
enhancements
Enabling switches • Many vendor specific
and routers to features still not
support centralized available as YANG
configuration models

• Standard configuration
not available across all
Configure vendors
YANG/network
configuration
• Developed CLI-based
models
adapters and Python
based APIs

8 8
 DevOps/Agile Framework
• Automating planning → Deployment
CI/CD pipeline
5 DevOps/Agile and testing
framework
Testing Automation Framework
Deployable Dashboard Monitoring
• Automating system, Interoperability and
Content
+
integration, Simulation testing across domains
TOSCA Docker images
Templates CI/CD with built code
Pipeline
Deployment Group • For CE 2.0 (E-Line, E-Access), L2/L3 VPN, Wave,
Code
Inspection
SD-WAN, vCPE.
Production Environment
Development
Environment Code
Repository

E.g. Challenges Mitigated

• Change requests approval
Module
Unit Test
Platform
Test
Staging Acceptance
Test
• Policy constraints for code change in
production
9
6 Enabling closed loop assurance What we did? E.g. Challenges mitigated
leveraging AI/ML
Big data lake • Skills pivot for big data lakes,
implementatio analytics, AI and ML
E.g. Performance Machine Learning Predicts events of different
Management Algorithm severities based on the n
(PM) Data intelligence built
• Custom ingestion code i.e.
Prediction integration various data sources
Custom graph such as alerts, performance,
ticket
visualization
Event Not an event

Critical Info Clear

• Visualization tool development
Data prediction & mail integration
PoC & porting
into production
• Model selection & training

10
 What we did?

7 Applications Operation and

Maintenance
• Proactive operation and maintenance
ensuring application availability

• Proactive application upgrade and

Decouple the
core and Backup the
Ensure enhancement based on product rollout
network
service declarative firmware and
management configurations software level
layer

Perform a Execute
diligent upgrade or Dry-run with E.g. Challenges Mitigated
canary testing patch decoupled
for all the enhancement system
components
• Seamless multi-vendor, multi-domain
production environment upgrade
Start and
couple the System
service Testing • Network traffic and application load
management
layer balancing
11
 What we did?

8 Managed Network Upgrade –

Enabling SDN Features
• Automated bulk firmware and software
upgrade, enabling SDN features

• CLI-based automated bulk configuration

across managed routers via orchestrator

Validate
Login to
managed
Configure
manageme Extract
Detect and
Trigger Publish
E.g. Challenges mitigated
syntax and match
device and nt inventory upgrade birth
configurati firmware
enable interfaces and upload and reboot certificate
on version
• Heterogenous network elements and
uplink ports and VLANs

software levels

• Multiple legacy NMS and EMS systems

12
 Service
Southbound Northbound
Templates
Interfaces (SBI) Interfaces (NBI)
(TOSCA/HEAT)
1 2 3
To enable E2E Service
Orchestration in Hybrid Routing and DevOps and Closed Loop
Networks Switching Testing Assurance with
Enhancements Automation AI/ML
4 5 6

Applications Managed
Open Source
What We Did Operation and
Maintenance
Network
Upgrade
Solution PoCs
7 8 9

Solution Platform Advisory

13
10
Key Challenges …Mitigated
• IT and network skillset combination • Partnerships (Prodapt, OEM vendors),
Skills pivot & Cross-trainings
• Legacy network programmability
• Plan-prioritize relevant user stories
• Moving to agile and DevOps
• Automated key aspects of CI/CD/CT
• Virtualization adoption • Early wins on agility and TTM for
service delivery
• Network automation
• Business and architecture ongoing • Vision alignment and management
involvement sponsorship

14
 01 02
Model-driven API-driven Automated CRUD
Configuration Templates Automated service creation based
on user configurations via APIs
Network abstraction based on YANG Model
for templatized configuration
Only specify CREATE operations,
REDEPLOY, UPDATE, DELETE
Key Learnings auto-generate

From ongoing
deployments
03 04
Transactional Guarantees Enabling Self-healing Network
Reduced manual intervention with templates Data Analytics systems and DSP Data Lakes can be
configuration and automated CRUD integrated for telemetry forecasting and analysis

Ensure fail-safe operations Enables error forecasting and demand patterns

15
 Benefits to our customers
Via an automated provisioning in multi-vendor, multi-domain ecosystem
in conjuction with our customer ecosystem

Reduced time
Faster revenue & CAPEX/OPEX Better customer
to rollout new
New revenue reduction experience
services

16
 Contact Us
THANK YOU
www.prodapt.com

www.linkedin.com/company/prodapt

www.twitter.com/prodapt

[email protected]

www.windstream.com

www.linkedin.com/company/windstream

www.twitter.com/windstream

[email protected]