Post on 10-Mar-2020
Openstack Summit 2018
Sana Tariq (Ph.D.) – TELUS Communication
AI Driven Orchestration, Challenges & Opportunities
AI/ML Driven Orchestration
Closed Loop Orchestration and Dynamic Policy
Service Orchestration Operational Challenges
Service Orchestration Journey
Agenda
NFV and Orchestration Journey…
Q2 2018
2016 2017 2018 2019 2020
IoT and OTT Services/ User-defined Services/International customers
2021 2022
PNFs Building NFV Cloud
Applications Onboarding
Increased Maturity OSS/BSS interlock, evolution of customer Portals, inventory compliance
Automated Assurance
AI driven/managed operations, capacity and applications
5G IoT customers defined services through customized user portals
AI driven/managed services (advanced)
Increased Maturity of Catalogs/Templates/Blueprints to deliver software defined services
NFV Telco Cloud - is different …
NFV Telco Cloud
APIs
Reliability
Self serve
Automation
Scalability
Standards
based
SecureCost efficient24x7
availability
High Throughput
Low Latency
Fast and
agile
SDN
Controllers
DC WAN
OSS/BSS
Ticketing Alarms AnalyticsBillingEnd to End Orchestrator
VIM
EMS
EMS 1
VNFs
VNF 1 VNF 2 VNF 3 VNF n
NFVI
EMS n
WAN
VNFMsVNFMs Service
Assurance
EMS 2
NFV and Orchestration Journey…
• Building robust cloud infrastructure
• Virtualizing Applications
• Provisioning workflows
• Assurance
• Automated scaling
• Traffic steered through SDN network
• Monitoring through holistic service assurance
• Integration with OSS/BSS
• E2E Service Orchestration plays are major “orchestration” role
Vendor lock-in/proprietary plugins
Depends on Company size
R&D driven roadmaps
Higher licensing cost
Trusted support model
Vendor agnostic/shared community plugins
Depends on community participation
Community driver roadmaps
Significantly lower costs
Self/Community support
Commercial Opensource
Opensource SI
Vendors
Orchestration: Commercial or Open Source
Orchestration Functional Elements
SERV
ICE
OR
CH
ESTRA
TION
SUC
CESS
CloudRobust Cloud supporting automation features, APIs, elasticity etc.
SDN Network Complete softwarization of DC and WAN for on-demand creation of services
AnalyticsRobust analytics cross-functional domains for scaling, healing & optimization for cloud and services
DevOpsEffective DevOps culture and support for short time-to-market and efficiency targets
Data ModelsHomogeneity across stacks for consistent data models for seamless integration, reusability and abstraction
Chaos of Multi-Vendor Multi-Domain…
S-GW P-GW
MME
vEPC
...
SD-WAN
vIMS
SBCS-CSCF
TAS I-CSCF
...
Enterprise
NFV Cloud
NaaS/OTT Voice OTT 5G IoT VoD
E2E Service
Orchestration
Customer facing services
Software Defined Service Operational Challenges…
Package and distribution
3
Orchestration (runtime)
4
Service Design
1
Troubleshooting
6
Assurance
5
Orchestration(testing in a sandbox)
2
Launch
Offline in lab Services Operation in real-time
Debug
Service failure
Policy Engine
Analytics
BigData/Hadoop
E2E Service Orchestration
LB
vFW vFW
Service Orchestration
VNFM
VAS
Collector
Closed Loop Orchestration Static…
Closed Loop Orchestration AI/ML…
BigData/
Hadoop
Inventory
WorkloadsOnboarding
Resource Orchestrator
Testing & Validation
APIs Adaptors SDK
Life Cycle Management
Policy
Service Definition
Workflows
Service Orchestrator Analytics
Conditional Rules
Cloud Services Security Network
Mapped Actions
Predictive/proactive Inputs
Reactive Inputs
AI/ML Orchestration
• Energy optimization
• Capacity optimization
• Faults prediction & healing
• Fast troubleshooting
• Traffic optimization SDN
controller
• QoS based routing
• Proactive/predictive threat
identification
• Closed loop decisions to
attacks mitigation
• Closer to the edge
• Service Healing
• Service optimization
• Differentiated QoS
Customer Experience
Cloud
Optimization
Network
Optimization
• 5G Services/ IoT
increase in traffic
volume and patterns
Data Surge
• Reactive decisions
• False Spikes
• Too many policies
& conflicts
Static Policies
Security
• Strategy to develop
dynamic policies and
testing
Dynamic Policy
• Robust Analytics
framework for
feeding AI/ML
systems
Analytics Model
WH
Y?
HO
W?
Wh
at?
START
PROCESS
STEP 1Problem definition
STEP 2Gather real, sizeable training data
STEP 3Prepping the data: cleansing, normalization, feature engineering
STEP 4Training phase: robust training environment
STEP 5Testing phase: Experiment with Models, Pick best Models and create feedback Loop
STEP 6Evaluation, predictor improvement and re-training
RESULT
Building an AI Application
?
Add your own description here. Add instructions, or additional information about this slide here. You can delete this item or any item on the slide.AI is becoming easier…
Data Scientists
needed, higher
complexity
Python libraries –
lower complexity
ONAP leverages
SPARK, Mlib,
MALLET, WEKA
etc.
H2O, TensorFlow Predictors &
Models developed
by shared
community effort +
data
Raw AI/ML ML Libraries AI Projects Acumos
• Better Analytics Engines,
• Better data
organization
• Large volumes of
managed Data
Computing Power
• Low Cost
• High Power
• GPU, NPU
availability
Accessible Data
ONAP Intelligent Closed Loop Architecture…
Design Time
SDC (Service Design Creation)
CLAMP (Closed Loop Automation
and Management Platform)
GUI that generates TOSCA
Run-time
DM
AA
P
POLICY EXECUTION ENGINEDCAE
Policy Execution
Hadoop
Synthetic Data
(Extension)
Ceilometer collector
Service Orchestration
SDN-Controller APP-C (GVNFM)
CeilometerCeilometer
AI/ML
(Python
Library)
TOSCA Policy1TOSCA Policy1
Condition
Cloud Region1
CPU Utilization
>80%
Action
Shift VNF/
workload from
Cloud Region 1 to
Cloud Region2
Policy1 Policy1
VNF1 VNF2
Cloud Region 1 Cloud Region 2
ONAP Policy Execution Flow…
Ceilometer
Analytics
AI MS Policy MS
Ceilometer Collector
Hadoop Cluster
DCAEDCAE
DM
AA
P
POLICY EXECUTION ENGINE
Service Orchestration
SDN-
Controller
APP-C
(GVNFM)
SDC
CLAMP
Cloud Region 1 Cloud Region 2
1
2
3
4
6
7
8
9 10 11
12
5
13
14
15
AI/ML Orchestration Industry Verticals
Study of VNFs performance requirements (MME, S/P-GW, vSBCs etc.)
NSD and VNFD Package description to associate color tags for VNF types to support these use-cases
DCAE: Dynamic Policy and Cloud optimization use-cases
Develop ML Models for cloud optimization use-cases
Enhance ceilometer & support cloud optimization configurations
BigData/
Hadoop/
Analytics
Customer/Resource Facing Portal
Customer with IoT
Traffic
(Dependency:
Closer to Edge
Location)
API Mediation
Inventory
WorkloadsOnboarding
Resource Orchestrator
Testing & Validation
APIs Adaptors SDK
Life Cycle Management
Policy
Inventory
Service Definition
Workflows
Service Orchestrator Policy Rules· Initial
Placement
· Cloud
Optimization
related
actions
· Customer QoS
related
actions
NFV Cloud
Customer with
real-time Voice
Traffic
Dependency:
Performance/ TTR
Customer with 5G
Services: real time,
high bandwidth
Dependency:
Performance/ TTR
Customer with
real-on-demand
viewing
Dependency:
Performance/ TTR/
throughput
Customer with off-
line downloading
Dependency:
Throughput only
Edge POD: High
Bandwidth, Closer to
customer
Zone1 PODs
Optimized for
Efficiency/
throughput
Zone2 PODs
Optimized for Low
Latency/HTTR (real-
time traffic)
Zone3 PODs
Optimized for Low
Latency/HTTR (real-
time traffic)
Sana Tariq (Ph. D.)
Sana.tariq@telus.com