BRKSPG 2402

The SP Service Edge
Transformation
with 5G, CO Transformation, Virtualization &
Edge Computing
Brian Meaney Distinguished Systems Architect
BRKSPG-2402
Cisco Webex Teams
Questions?
Use Cisco Webex Teams to chat
with the speaker after the session
How
1 Find this session in the Cisco Events Mobile App
2 Click “Join the Discussion”
3 Install Webex Teams or go directly to the team space
4 Enter messages/questions in the team space
BRKSPG-2402 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 3
Service Edge - Many Perspectives !!!
Public Cloud
Centralized DCs / Clouds
SP Edge Cloud IOT Edge Cloud Ent Edge Cloud Public Cloud Providers @Edge
SP Multi-access Edge IoT Edge Enterprise Edge Cloud Services Edge
Edge on the Devices

Agenda
• Market Trends, Analysts Viewpoints

• Industry Initiatives & Technologies
• 5G Evolution & Innovations
• Service Edge Requirements
• Cisco Service Edge and the
Distributed Telco Platform
• Key Takeaways
Market Trends
(Influencing Service Edge)
The SP Market Is Subject to Five Megatrends
Explosive growth Advent of 5G Changing Rise of cloud Cyber security

of video and subscriber user and web scale
mobile experience players
• Traffic growth driven • Service • Integration across • Majority of traffic • Widening attack
by video (>80% in awareness and multiple networks to originate from a surfaces and
2022) and unlimited enablement small number of increasing rate of
plans • Self-service / content providers attacks
• Network control /
• SPs’ networks to assurance, immediacy • Emerging • Multiplication of
evolve into cross- policy, and SLA Distributed Cloud IoT endpoints /
medium, converged for Enterprises requirements DDoS
delivery networks
Customer & App Experience is Taking Centre Stage
Mobile ARPU, Multiple Countries
50
45
40
35
30
25
20
15 Source: EU Commission
10
5
0
2015 2016 2017 2018 2019 2020
Consumer ARPUs are Declining or Flat B2B or B2B2x Market Has Future Growth
Low Latency for better QOE and to

Enable New Applications, Customer
Experience Transformation
Software Disrupting Every Domain
Mobile Packet Core (MPC)
• World’s Largest vEPC deployment with Cisco
Cloud Radio Access Network (CRAN)
• Cloud-Native 5G Core
• xRAN/O-RAN Alliance – SP led initiative for
virtualization, disagg & multi-vendor RAN • 4G/5G CUPS – UPF optimized with VPP
Cisco co-author of xRAN drafts • MPC and Transport convergence with SRv6 –
NETCONG/YANG defined for RAN integrated service chaining/network slicing
• Open vRAN ecosystem - accelerate adoption of

multi-vendor vRAN (xRAN/O-RAN,TIP OpenRAN) virtual CMTS
Led by Cisco with industry partners, namely, • Cloud Native – unique solution in the market
Intel, RedHat and several RAN vendors • Open APIs – REST API
• Streaming Telemetry
• Continuous integration and continuous
development (CI/CD)
CMTS - Cable Management Termination System
Service and Transport Decoupling
Unconstrained by the transport, multiple service overlays are delivered over the top
Centralised Centralised
Service Control Service Control
Service
Overlay
Centralised
Transport Policy
and Optimisation
Hybrid SDN model
= Service aware
Transport
Underlay = Service unaware
Distributed Connectivity Control
As the Service Evolved….Workloads Evolve……
App App App App App App App App App
Runtime Runtime Runtime Runtime Runtime Runtime Runtime Runtime
OS OS OS
Host OS VM VM VM Host OS Host OS
H.OS / Hypervisor H.OS / Hypervisor
Hardware Hardware Hardware Hardware Hardware
Bare Metal / Dedicated Virtual Machine Containers in VM Containers (on BM) Function-as-a-Service
Worldwide Container Instances Installed Base by Virtualized/Non-

Virtualized (excluding web/SaaS provider internal infrastructure), 2015–
2020
1,000,000,000
900,000,000
800,000,000
700,000,000
600,000,000
500,000,000 Containers in VM
400,000,000
300,000,000
Containers on BM
200,000,000
100,000,000
0
2015 2016 2017 2018 2019 2020
Virtualized Non-Virtualized
Networked Devices, User & Service Growth Forces
Traditional operations staff will not be able to

72% increase keep up with adding ~10% extra devices every
year without changing how they work
Automation can provide that catalyst for change
Western Europe
Service Edge - Analysts Viewpoint?
“Over the Last Two Years, IT Organizations Spent “By 2021, at least 75% of large and global enterprises
70% on “Run the Business” IT Spending, up from will implement a multicloud-capable hybrid integration
67% in 2013 to 2014, & 65% in 2012” platform, up from less than 25% in 2018”
2018 Gartner, Strategic Roadmap for Compute infrastructure 2019 Gartner, Strategic Roadmap for Compute Infrastructure
Traditional Infra IaaS/PaaS/SaaS
Private Cloud Private Cloud
Simplify Multi-Cloud
Scale-Out Infra Scale, Platform & Consume, Connect Public Cloud 1
E2E Orchestration & Protect
Mgmt. and Ops Public Cloud 2

Edge
MEC, Service Edge,
CO Transformation
“In 2018, 85% of operator respondents plan to “By 2022, more than 50% of enterprise-
create, or will have already deployed, smart central generated data will be created and processed
offices … installing servers, storage and switching to IoT Branch/ outside the [central] data center or cloud”
create mini data centers in selected central offices” Remote
Real Time
2018 IHS Markit Operator Survey 2019 Gartner, Strategic Roadmap for Compute Infrastructure
Analytics and
Inference 13
BRKSPG-2402 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public
Industry Initiatives &
Technologies
Standards & Industry Group Landscape* * Partial List
Quite Fragmented at the Moment – Needs focus
Reference
SA2 - Platform /
MEC WG MEC WG Fast Data Plane Implementatio Many Topics Orchestration
Architecture NFVI
n
Disaggregated Just started WG Defined MEC Cloud Native Fast Data Plane Open Many topics: Defining open
mobile core looking to Release 1 Platform For for Cloud Native Reference - In-band architecture &
architecture establish MNO architecture/API NFV & Edge VNFs and implementation meta-data Implementatio
(CUPS) and implementation and Now Computing Infrastructure and Validation comm. n for MANO
URLLC guidelines focused on R2 - hICN, QUIC
Architecture & Edge NFV, Edge Virtualization Fast Virtual NFV Many Topics Orchestration,
Service Edge Computing Computing Infrastructure Networking Automation
Industry Reference Project Reference

Many Topics Compute HW APIs Ecosystem
Collaboration Implementation Group Architecture
Foster industry Bring Cloud Many Topics: Lab and field Edge Defining APIs Defines ref. Provide ref.
coordination in Agility + DC DSRC, implementation Computing and common architecture implementatio
an open and Economies to 6LowPAN / RPL of edge in any Hardware capabilities to for Fog n, do live
innovative way Transform CO & etc.…. access enable M2M Computing (7- demo at test
Service Delivery service layer pillars) and trial center
Edge CO Many Topics Infrastructure Infrastructure Edge Edge Edge
Computing Transformation Computing Computing Computing
SP Virtualisation Landscape aka NFV/Telco Cloud
1 Do it Yourself (DIY) 2 Vertical Solution Stack 3 Common Horizontal Platform
MANO MANO MANO MANO Common MANO
VNF VNF VNF VNF VNF VNF VNF VNF VNF VNF VNF VNF
Virtualization Virtualization Virtualization Virtualization Common VIM / Virtualization
Hardware Hardware Hardware Hardware Hardware Hardware Hardware
• NFV started here • Many operators are here • Most SPs want to achieve this
• Mostly lost traction except a • Essentially an appliance like • Some did this partially
handful of operators model. Multiple vertical silo’s, • Multi-vendor interoperability was
cited as the biggest challenge
Majority of the market today belongs Here
SP’s want to get Here

NFV – Network Functions Virtualisation
The Market is taking off….
• Common Platform with Cisco VIM (UCS based) • Vertical stack for Cisco SP Mobility VNFs
• ACI based network underlay • Cisco UCS, Nexus, RHOSP based (CVIM TBD)
• Cisco Ultra vEPC, PCRF, 3rd Party Gi VNFs, IT apps next • VNFs – Ultra S/PGW, PCRF
• ~15 sites in production, close to 1Tbps BW & 2Mn subs • 46 numbers of location, 79Mn subs in production
• 72 hours from site delivery to first data call • CUPS targeted in 2019, eventually up to 2000 locations
• NFVI with CVIM at 12 DT Core POPs • First true realization of distributed telco cloud
• Core POPs are being used as Edge • Common Platform with Cisco VIM and Cisco MANO
(ESC & NSO) Software, ACI & Nexus switches
• 20ms RTT engineered with 12 POPs
• Fully virtualized – Cisco Ultra vEPC, Altiostar vRAN,
• MobiledgeX – (Aggregator) is the first customer
Nokia & Mavenir vIMS, ASAv, NEC OSS/BSS…...
• Plan to Extend to 100s of more sites for Service Edge
• ~300 sites targeted in CY2019, Eventually ~3000
CO Transformation (or CORD)
In reality, the goal is to create an Open Platform for Service Migration & Creation
Transformation of the Network Architecture by applying NFV, programmability, SDN and
DC-like design; to help bring economics of DC and Agility of Cloud at the CO
NFV and SDN DC Architecture Cloud Agility
CO - Central Office CORD – Central Office Re-architected as a DC

Multi-Access Edge Computing (MEC)
MEC or Edge Computing, is an architecture principle of moving services closer to the user, at
the edge of a network to enhance experience and enable delivery of low latency apps
Latency Reduction Data Reduction Offload at Edge
~ 1 – 30 ms ~ 100 – 200 ms Anchor Server

5.34
Edge Server
Edge Server
5.34
Cluster
Reduced Bandwidth
Host
Edge Server
Host
Reducing latency between services Edge nodes can perform data analytics Edge offload will enable less expensive
and consumers will create a better (ML inference) to perform bandwidth and lower latency path from the edge
QoE & allow for new B2B2X reduction and/or compute offload hosts towards the services
services compensating for less capable devices
Edge Transformation - Deployment Models for SPs
Premises-Based Network-Based
End
QoS/Service Control/ Customer App App
Control
PCF
Policy
App App Services
SMF
App App Infra workloads UPF

PCF
App App SMF
Services Access Platform Platform
Services
Infra workloads UPF
Platform Access Platform Network Network
Services Macro Network Edge Core
Enterprise Location Premises

Enterprise/Customer Network
• Places the edge workloads on the customer • The edge workloads are at a low latency
premises location location with respect to devices (a CO)
• Use cases are enterprise: factory • Use cases can be consumer or enterprise
automation, medical, corporate campuses & but initially are infrastructure based
require domain expertise • Opportunities leverage the reach of the
• Private radio (licensed or unlicensed) is a macro network
significant part of the operator opportunity
“Dedicated MEC” “Distributed MEC”
Service Edge Transformations
Central Office Transformation Service Edge Transformation
NFV and SDN DC Architecture & Cloud Agility

Economics Virtualization Decomposition Placement
Transformation of the CO Architecture (CORD) by applying Centralized, Appliance Based Mobile Gateway
NFV, SDN and DC-like design; economics of DC & Agility of
Cloud PGW
eNB
Edge Computing Cell Site CRAN Hub/ CO / Agg. / Regional DC Central DC

Pre- Agg. MSO / HE
Distributed, Virtualized,
Disaggregated, Decomposed
Virtualized, DecomposedMobile
RAN &Gateway
Gateway
vPGW-
Latency Edge Data vPGW-
CPF
UPF
Reduction Offload Reduction RRH
vCU
MEC is an architecture moving services closer to the user to the vDU

edge to enhance experience & enable delivery of low latency apps
Cell Site CRAN Hub CO / Agg. / Regional DC Central DC
/Pre- Agg. MSO / HE
MEC – Multi-Access Edge Compute CORD – Central Office re-architected at a DC
Terminology for Service Edge
Public Cloud
Provider
Location Type → Cell Site / C-RAN Hub / CO / Agg / MSO / HE Regional DC Central DC
Access Pre-Agg.
10,000’s 100s - 1000s 10’s / 100’s 10+ to Few 10’s (<10)
Terminology from a Services Edge or Applications Placement Perspective
(And not just a function placement perspective → decouple the function vs services placement)
Terminology
Far Edge Edge Regional DC Central DC Public Cloud
we’re using
EMEAR Frontend Data Backend Data Center

Customer SP #1 Center
Terminology EMEAER Metro/Aggregation Regional DC Core DC
SP #2
Distributed Telco Cloud

A virtualized telecom infrastructure, distributed from Central Data Centers towards the access with a
consistent management framework, running any telco apps for end-user services with agile operations
5G Evolution & Innovations
5G Key Use Case Categories
Increased Bandwidth
Enhanced Mobile Broadband (inc. Fixed Access) and Capacity
• Extra capacity delivered through new 5G frequency bands
• Not too concerned with connection density or latency.
Slicing, Flexible deployment,

IoT/massive Machine Type Communications NFV/Virtualisation
• Focused on low power wide area NB-IoT with high
connection density and energy efficiency
Push data plane to the edge,

Ultra-Reliable Low Latency Intelligent in Network
• For mission critical use cases (self driving, Public safety, ...)
• Desired 1ms access time only refers to radio interface and
would be most useful in near field mission critical apps
Source: Recommendation ITU-R M.2083 TR 38.913 specifically references vehicle-to-everything (V2X) requirements as a defined use-case
5G Decomposition & Disaggregation AUSF - Authentication Server Function
UDM - Unified Data Management
AMF - Access & Mobility MGMT Function
N13 SMF - Session Management Function
AUSF UDM PCF - Policy Control Function
AF - Application Function
N8 N10
Control Plane N12
N11 N7 N5
AMF SMF PCF AF
N1 N14
N2 N15
N4
F2 F1
UE RU DU CU
Air
Radio E1
N3 UPF N6 DN
Xn
F2 F1 N9
Data Plane RU DU CU
RU – Radio Unit
Radio E1 DU – Distributed Unit
CU – Centralised Unit
UPF – User Plane Function
DN - Data Network
5G Architectural Shifts
CUPS and MEC Network Slicing Cloud RAN
What? Distributed workloads for Logical, virtually Functional split and

services and applications independent end-to-end Virtualization of Radio Access
networks Network
Why? • Services (Low latency, QoE) • Services, Monetization • Lower costs

• Lower transport costs • Differentiated SLAs • Simplify operations
Network • Any-to-Any connectivity • End2End SLAs, Steering • Low latency

Requirements • Smooth integration between • Dynamic and flexible • High throughput access
WAN and DC fabric service management
CUPS: Control/User Plane Separation

MEC: Multi-access Edge Compute
Decomposition of Network Functions - CUPS
• Decompose the Traditional gateways

Subscriber Access Platforms
into control plane and a multiplicity of Billing Policy
Manager Regulatory
user planes. IMS
• Control and User Plane Separation Access Access

Node
Subscriber
Platform Video
(CUPS) in 3GPP Internet
Sub DB
• Enable User Planes (UP)s to be

placed anywhere in the IP network, Conventional Subscriber Management
e.g. closer to the consumer on the Billing Policy Regulatory
Edge
Manager
• Augment UPs with with in-line

Control
Plane Sub DB
services IMS
Access Access
Node
User Plane
Function
WAN
• These types of functional
Video
disaggregation are enabled with Access Access

Node
User Plane
Function
Internet
virtualized workloads
Decomposed Subscriber Management
Cloud RAN/vRAN - Architecture Evolution
Distributed RAN Centralized RAN Cloud RAN/Virtual RAN
EPC/5GC EPC EPC/5GC
S1/NG
IP
Backhaul
Backhaul S1
IP vCU
F1
Midhaul IP
Backhaul BBU
S1/NG vDU
IP
eCPRI/
Fronthaul RoE
Fronthaul CPRI ORAN
TDM IP/Eth
RU RU
RRU RRU
RRU RRU
BBU BBU
4G/5G 4G 4G/5G
Traditional Deployment
Centralized baseband units Virtualize and Decompose BBU
BBU co-located with RRU (Full RAN
CPRI interconnect New functional split (CU/DU/RU)
stack @cell-site)
Not virtualized Packet based interfaces
Dedicated HW BBUs
RU – Radio Unit DU – Distributed Unit CU – Centralised Unit
RAN Functional Split Consideration
To Mobile core To Mobile core
To Mobile Core To Mobile Core

Split 8 Split 7-2x Split 2 Backhaul
RAN
RAN Stack
Stack
SDAP SDAP SDAP
RRM/RRC Option 1
Centralized Centralized
RRM/RRC Option 1 RRC Unit Unit
RRC PDCP RRC PDCP PDCP
PDCP Option 2
PDCP Can be
RLC High RLC RLC Option 3
Option 2
combined
Midhaul
High
Low RLC
RLC Option
MAC MAC MAC MAC
Option 4 3
Scheduler Scheduler
Distributed
High
LowMAC
RLC Option 5 4 Unit
Option
Hi-PHY Hi-PHY
Low MAC
High MAC Option 6
gNB
LO-PHY Option 5
CU High PHY Option 7 Fronthaul
Low MAC
DU gNB Low PHY Option 6
Option 8 RLC
High PHY (CPRI)
RU CU
RF Option 7
MAC MAC
5G NG
Scheduler
DU Low PHY
Option 8 (CPRI) RU DU DU
Hi-PHY
RU RU
RU
RF RU DU
LO-PHY LO-PHY Can be
5G combined
RU
NG RF RF RF
DU: Distributed Unit
RU: Remote Unit
Fronthaul bandwidth* and latency requirements (100uS transport allowance) are critical in determining position of (v)DU
*Heavily dependent on radio characteristics i.e. no. of carriers, Frequency band, channel size, modulation, no. of Antenna etc.
Cloud RAN/vRAN – EMEAR Deployment Options
Cell site Pre-Agg Aggregation Edge
vUPF D-RAN
Nx, ~user b/w, msecs delays
BACKHAUL
A
RU/DU/CU
F1: ~user b/w, msecs delay Nx: user b/w, msecs delay vUPF C-RAN
BACKHAUL
B MIDHAUL
RU/DU
vCU
eCPRI v.high b/w, µsec delay Nx, ~user b/w, msecs delays vUPF C-RAN
BACKHAUL
C FRONTHAUL
RU DU/vCU
eCPRI v.high b/w, µsec delay F1.Nx, b/w=user rates, msecs vUPF
C-RAN
delays BACKHAUL
D MIDHAUL
RU
DU vCU FRONTHAUL
Open RAN Industry Momentum
Service Provider Problem RAN economics no longer work.
Statement Must reduce costs and accelerate innovation
Industry Response Complimentary Operator and Vendor sponsored consortiums have been
created with goal of opening up RAN and shifting from HW to SW
• Operator Lead – Drive Open, Standard Interfaces

O-RAN • Bring virtualization, decomposition, disaggregation and multi-vendor to the RAN
Alliance • Cisco co-authoring standardization specs and demoing open Management,
Automation – NETCONF/YANG for RAN Management
TIP • Operator Lead – Intel providing major input

• Concerned most with the disaggregation of h/w and s/w
OpenRAN
• Vodafone & TEF active on OpenRAN, BT, DT, Facebook active in TIP
• Cisco Lead, Operator Driven – Compliment ORAN/XRAN

Open vRAN
Cisco Led • Accelerate Innovation – Software Centric, Forward Looking
Ecosystem • Drive Open/IP Architecture
Customer 5G Perspective of Network Slicing
Network Slicing is fundamentally an end-to-end partitioning Benefits
of the network resources and network functions so that • Each separable business operation can be
selected applications or services may run in isolation from efficiently and reliably run on a network slice
each other and for a specific business purpose • Alternate policy and charging structure
• Unique service assurance characteristics
API Programmability and BSS
• Increased service security
Slice Service Orchestration (NSO)
Enterprise
• Infrastructure orchestration manages the
NSSF AUSF
Sliced 5G Core Slice
Tenants
complexity driven by the requirements of
each slice
UDR
Video
client
PG
Sirgenn
StalAD
1gtBJ2h WE36
M
H47 CTK58 N
RG AN
AMF
Q XF9
R*I U L0 O Y#
S VZ
Wireline Slice Common

VPN
Leverages the SP distributed DCs and Footprint
Functions
•
Resource
• Each slice can have its own MANO/OSS
SMF
UPF
PCF AF
AF
Intra-Slice
Management Management environments
VPN AF Intra-Slice
• New service introductions are quicker

Management
RAN Slicing UPF Intra-Slice Data
Management Network
Data
UPF Svcs Data

Network
Network
• Slicing has a significant reduction in regression
VPN testing cycles
RAN Slicing Common Layer for Resources • Isolation eliminates effects of rogue applications
(E.g. M2M)
Physical Spine/
Smaller failure groups imply no single “too big to
Compute Storage
Resources Leaf
•
fail” node
E2E use-case
Access Domain
Network Slicing - Cisco E2E Domains Virtualised
Radio
(ORAN Partner)
Cross-Domain Automation & Orchestration

Smart
City/IOT 5G CN Core
Central DC
AR/VR/ mMTC Slice

Gaming
Access Aggregation Core

Internet eMBB Slice
Ent/Public Sector
Applications
Collaboration
Services Edge DC Regional DC
Private ENT uRLLC Slice
Core DC
Security
Services
Security Domain – Full-Stack Security
Mobile Core Domain Telco DC Domain Transport Domain Orchestration Domain

Telco DC & Transport with Cross-Domain
Mobile Packet
Virtualisation Segment Routing Automation &
Core Components
Infrastructure for Network Slices Orchestration
© 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public
Service Edge Requirements
(What is the Service &
Where is the Edge?)
Edge - Functions, Services & Use-cases
Partial List
Foundational – Providing Baseline Platform On-Demand
Infrastructure Use Cases Video, Live TV
Network to User
vRAN / Cloud RAN, CUPS/ Decomposed Mobile
Packet Core, CUPS based BNG, (v)OLT, vCMTS
with R-PHY, Gi-LAN Services etc.
Connected Vehicle
to Network, User
User to Network Generated Live TV
Operator Branded Services etc.
Content streaming via CDN, Live TV with CDN, IOT,
Fog Computing, AR/VR based services etc.
Localized AR, IOT,
Interactive Gaming,
Vehicle-to-Vehicle
User/Edge to User/Edge etc.
Services to Businesses
B2B Online Gaming, Connected Autonomous Vehicle,
AR/VR Services, IOT Services, Public Cloud
Hosting, 3rd Party App hosting, Edge Analytics Business Managed
Services
Bi-Directional (User – Network)
Locations, Current State and Questions
Connectivity Service: Packet
Managed Managed
Managed Core, Policy, IMS,
Access: SP Mobility is Service: Svces: vMS
Svces: Virtual Conn Optimization: CDN,
eNB/NodeB, Cloud Video: SP
Branch, SD- Pre-Dominant NAT, FW, Video/TCP Opt .. Services
OLT, CMTS Video CP/DP
WAN Managed Svces: VMS Interconnect SaaS Services
Video: SP Video CP/DP
Internet
Public Cloud
Provider
Cell Site/Access C-RAN Hub/Pre-Agg. CO/Agg/MSO/HE Regional DC Central DC Peering

Branch / ~10’s <10
Venues 1000’s/10000’s 100’s/1000’s 10s/100’s
What may be What may be What may be Will anything

deployed here ? deployed here ? deployed here ? change here? Will anything
change here ?
Who owns it? Who owns it? Who owns it? Who owns it?
To Distribute or Not to Distribute?
Advantages and Disadvantages of Distributed Deployment
Reduction of Latency Operational Complexity
(“Latency is Money”) (Perhaps the most important issue)
Reduction of Backhaul Higher Infrastructure Cost

(Fiber/Transport Owned vs Leased?) (Need to show the business case)
Location Awareness Location/Footprint Issues

(New Services Opportunity) (Typically not built for DC infrastructure)
Longer Battery Life Security Concerns

(For IOT Devices) (Including Physical Security)
Regulatory/Compliance Environmental Issues

(May apply more in multi-country ops) (Hardening mean more cost to build)
Localized Impact of Fault Technology Maturity

(Smaller blast radius) (Matter of time – we will get there)
Customer Survey “Where is that Edge”
Deep analysis conducted across Multiple Theatres on Top 6 Key SP use-cases
Virtual Gateways/App Cloud RAN Public Cloud Hosting Immersive Augmented Interactive Gaming Connected Autonomous
(vBNG, vCMTS, CUPS, CDN) (RU, DU, CU) (AWS GG) Reality Vehicle
“Where is the Sweet Spot ?”

Public Cloud
Provider
C-RAN Hub/
Pre-Agg. CO/Agg/MSO/HE Regional DC Central DC
Latency Increases in this Direction
Cost of Backhaul Increases in this Direction
Operational Complexity Increases in this Direction
Cost of Overall Telco Cloud Infrastructure Likely to Increase in this Direction
Use Cases for Edge
(Open) Virtualized RAN Decomposed Mobile Core (CUPS) vCMTS with Remote PHY
SP Infrastructure
Control Plane
CP
RU
Distributed
Unit
Service
UP
Centralized
Unit
UP Core Edge Services
Edge
RU Distributed
RAN UP
Unit
Edge Services IMS
RU
Edge CDN for (Mobile) Video Interactive/VR

Business (B2B) Use Cases for Edge
Gaming Gaming Catalog

Operator Branded or Business to
Node
Edge
Cache APIs
GW + Gi
Mobile Network Mid-Tier
Cache
Virtualization
Match
Platform Maker
Edge
Cache
CO / Agg / Public Cloud
MSO / HE
Public Cloud Connected

Hosting Public Autonomous Ent. Edge
Cloud Apps Vehicle (Big Data) Cloud
BM- aaS ML / AI
GW+Gi APIs
UPF + Gi
Virtualization Platform
RU DU CU
CAV Apps on
Ent DC or
Cloud
Enterprise © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public
Outcome #1 - Service Edge Requires an Ecosystem
Business
{...} Use Case:
Developers, Edge Cloud
Wholesale
Enterprises, SaaS
IAAS Provider Providers Consumer
System
Integrators
Fixed
Use Case:
Virtual CDN
CDN Provider Content Owners, Providers Consumers
Component
Vendors Mobile Use Case:
Connected
MEC Vehicle
Car
Distributors Car owners
Suppliers OPERATOR Manufacturer
Source: STL Partners + Cisco Changes
Outcome #2 - Mobile Network Transformation
CRAN/vRAN and EPC evolution adoption will be a defining decision
Today
eNB IMS / vIMS
Backhaul IP Core
EPC / vEPC
eNB
S1
Transformation Underway
Stringent Latency requirements 250us/100us UPF
MEC
Apps
N3
eCPRI (v)DU F1 vCU
RU vIMS
Backhaul / IP
Fronthaul Midhaul Core vPC-CP or
vEPC
RU
Status Quo Architecture evolution has potential to Adjacent evolution in RAN has potential to
impact our transport business impact Packet Core as its evolving with
CUPS
EPC – Evolved Packet Core RU – Remote Unit DU – Distributed Unit CU – Central©Unit

2020 UPF
Cisco and/or its – User
affiliates. Plane
All rights Function
reserved. Cisco Public
Outcome #3 - Application Requirements
Latency plays a major role in the location of the service
Use Case Latency Notes
Mobile Video ~75ms (One way Including ~25ms of buffering. This recommendation is derived from considering typical 𝑃𝐿𝑅 in radio and
Delay), will its impact to the performance of TCP congestion control. It implies maximum OWD between streaming
depend on PLR server and RAN node must be < 35 𝑚𝑠 for LTE and < 40 𝑚𝑠 for 5G.
Mobile Augmented Reality 10ms (One Way Network-based inference will not result in viable mobile AR in LTE since the air interface budget (30 𝑚𝑠)
Delay) exhausts the smallest possible lag (20 𝑚𝑠). In 5G, the AR inference capability likely requires an edge host
adjacent to the RAN node.
Mobile Virtual Reality, 20ms (One Way), For VR, Based on a 40 𝑚𝑠 RTT requirement for maximum lag, we see the maximum One Way Delay
Interactive Gaming 50ms (One Way) between the VR server and the RAN node must be < 5 𝑚𝑠 for LTE and < 10 𝑚𝑠 for 5G.
VoIP 200ms (One Way Note that the number implies VoIP can be supported over longer distances with no measurable impact on
Delay) delay. This suggests IMS APN should remain centralized
Other Low Latency / uRLLC Use Cases (May be deployed more in Future )
Factory Automation 0.25 − 10 𝑚𝑠 Real-time control of machines and systems in production lines
Intelligent Transportation 0 − 100 𝑚𝑠 Autonomous driving and optimization of road traffic (platooning and overtaking)
Robotics and 10 − 100 𝑚𝑠 Remote control with synchronous visual-haptic feedback

telepresence
Health care 1 − 10 𝑚𝑠 Medial Bio-Telemetry, Tele-diagnosis, tele-surgery
Smart Grid 100 𝑚𝑠 Switching on/off electrical sources to compensate for demand fluctuations
Fronthaul latency requirements (Infrastructure Service) are critical in determining position of (v)DU - 100uS
Finding a sweet spot for the Service Edge
Current Service Requirements indicate a Sweet Spot Latency of 10-20ms (user to App)
May be in future (Pre-Agg / C-RAN Hub)
For uRLLC requiring <10 or <5ms e2e OWD Sweet Spot (Agg/CO)
5G-NR air i/f delay improvement over time can help to retain @ CO to Enable Initial Services
CSR to Pre-Agg is hub-&-spoke connectivity, distance of ~10KM.

Assumes 8 Pre-Agg + 2 Agg devices in each aggregation ring of total distance of 100KM.
2 ms 2-4 ms 2-4 ms CSR to Pre-Agg = 10us + 20us + 50us (10KM fiber transmission) = <100us
15ms(D)/20ms(U) Pre-Agg to Agg Ring = 8 x 20us + 500us (100KM fiber transmission) = <1ms
in LTE < 1 ms
~100us 2ms Device (UE) processing delay, 2-4ms compute/app processing delay
0.5 – 8ms (5G)
Net-net: ~30ms uplink & ~25ms downlink for LTE; ~20ms for 5G
C.DC Peering
Pre-Agg / Agg / CO PE Core
C-RAN Hub / MSO
CSR
2000-5000X 200-300X 15-30X 3-10X 1-5X X Location #
• Modeling End-to-End Latency is key – User to APP, it is not only transmission or air interface latency
• On the IP transport infrastructure, queuing is a significant contributor, which Cisco solution can optimize
Places in the Network – Where is the Edge
Latency | Cost of Backhaul
Operational Complexity | Cost of Infrastructure | Economy of Scale
10-20 ms
~100us
Internet
Cell Site
Public Cloud
Provider
Managed CPE C-RAN Hub / Pre-Agg. CO / Agg / MSO / HE Regional DC Central DC Peering Points
10000’s 1000’s 100’s 10's <10 ~10
Non SP facilities Location Proper DC Facilities.

Constrained
Constraints Focus of Present
Locations
Varies NFV Deployments
Remote UPF for Only viable Becoming Sweet

specific Enterprise location for Spot for Edge
Use cases e.g. ULL vBBU or vDU Computing Apps &
(based on 4G Services
requirements)
Next Phase
What May likely to go where?
Connectivity Connectivity
Service: EPC Connectivity Service: vRAN (vCU), Conn. Service:
Service: vRAN CUPS-User Plane for Connectivity Service: Packet Core,
CUPS (vDU, vBBU)
Edge Mobile/Wireline Gateways, vCMTS Control Plane of Policy
Computing: CUPS Gateways Managed Svces:
For some SP’s:
CDN, IOT, Edge Computing: Analytics, IOT, CPE / SD-WAN,
EPC /BNG CUPS,
Localized AR, vCMTS, Edge AR/VR/Gaming, Public Cloud FW
Analytics services Hosting, CDN… SaaS Services
Happening Now
Connectivity Service: Packet Managed Managed

Managed Core, Policy, IMS, Conn Service:
Svces: vMS
Svces: Virtual Optimization: CDN, NAT, Cloud
Services
Video: SP
Branch, SD- FW, Video/TCP Opt ..
Interconn Video CP/DP
WAN Managed Svces: VMS
ect SaaS Services
Video: SP Video CP/DP
Internet
Public Cloud
Provider
Branch / Cell Site / C-RAN Hub / CO / Agg / Regional DC Central DC Peering

Venues Access Pre-Agg. MSO / HE
Cisco Service Edge and the
Distributed Telco Cloud
Platform
Telco Cloud – Overall Requirements
Cloud Infrastructure Network Services Operations Model
• Virtualisation Platform • Centralized vs. distributed • Lifecycle Management

• Implementation: VM, Container • VNF: control and/or user plane • Interoperability & Open API
• Performance: IT vs. Telco • DC Fabric and Compute • Orchestration and Assurance
• Evolution to Unified Cloud • Intra & Inter-DC service chain • Tools, processes and skillsets
Telco Cloud is moving to the Edge…
Typically 30-50+ ms (Round Trip)
10-20ms (Round Trip)
~1ms ~100µs
Edge
Computing:
EPC CUPS, Connectivity Service:
CDN, IOT, Connectivity Service: vRAN, Gateways,
AR, Analytics vRAN (vDU, vBBU)
Edge Computing:
For some SP’s: EPC AR/VR/Gaming,
/BNG CUPS, Edge Connectivity Service: EPC, Policy, IMS Managed Managed Svces:
Public Cloud Hosting,
Managed services Conn Optimization: CDN, NAT, FW, vMS
CDN, Analytics, IOT Service:
Svc’s: Virtual TCP Opt .. Cloud Video: SP Video
Branch, SD- Managed Svces: Video: SP Video Services CP/DP
WAN CP/DP Interconnect SaaS Services
Internet
Public Cloud
What will be Deployed here? Provider
Branch / Cell Site Pre-Aggregation Aggregation Regional DC Central DC Peering

Venues 10000’s 1000’s 100’s 10’s <10
Next Phase with 5G & MEC Current Phase with 4G
Cisco Telco Cloud Stack Composition Today
Cisco Orchestration Values NFVO – Cisco Network Services Orchestrator (NSO)

Model Driven Orchestration (Model Driven Multi-Vendor Orchestrator)
Open & diverse set of interfaces
Rich multi-vendor support
NFV & Resource Orchestration VNFM – Cisco Elastic Services Controller (ESC)
Security and Assurance

Flexible & Scalable (Generic VNF-Manager)
Any Virtualized Network Function

(Cisco or 3rd Party)
Cisco VIM Values Cisco Virtualized Infrastructure Manager (CVIM)

Fully Validated
E2E Automation & ZTP (Cloud Orchestration (Openstack) with Virtual Compute, Storage & Network)
Carrier Grade and RT capability
Platform Security
Open and Modular
Lifecycle Management
Compute and Storage Networking
(Cisco UCS or 3rd Party) (Cisco ACI, Nexus 9000, NCS5500)
Rich Operational Tooling
Cisco Telco Cloud Stack Composition - Partners
Automation Framework: Cisco BPA, CDAF
NFVO, Resource Orchestration & VNF Service Orchestration
Cisco NSO, Cisco Cloud Center Suite Service Assurance: Situation Manager, Matrix
Virtual Network Functions (Cisco and 3rd Party) VNF Manager

Connectivity Service Business Managed Connectivity Telco and Admin Cisco ESC 3rd Party
Optimization services services Applications G-VNFM VNFM(*)
Virtual Infrastructure Cisco VIM

Infrastructure Management
Kubernetes Cluster Cisco Container

API
Unified Management
Platform
NFVI Assurance
Virtual Compute Virtual Storage Virtual Network OpenStack API

(RHEL) (Ceph) (Fd.io, SR-IOV, OVS)
Cisco VIM
GUI Lifecycle Manager
Infrastructure Abstraction with RHEL, KVM/Qemu, Host Packages, vSwitches
Physical Infrastructure Network VIM (opt.)

Compute Network Storage
Cisco UCS NX9K/NCS5500 Cisco
Cisco NFVI (Network Functions Virtualisation Infrastructure) Scope (*) ETSI NFV SOL001 & SOL003 Compliant
Software Defined 5G
Open vRAN & Edge Computing Architecture
Zero Touch, End-to-End Automation with NSO, ESC
Small Cell
The Internet
vRAN IPv6 based 5G Ready Transport Network for Backhaul
4G/5G Rakuten
Hybrid Cloud
~3000 Edge DCs ~50 Regional DCs 2 Central DCs
vCU MEC App vIMS
vDU UPF vEPC
CVIM CVIM CVIM
Nexus ACI ACI
Distributed and Common Carrier-Grade Telco Cloud
Software Defined Programmable Infrastructure
Open, Decomposed, and Edge Computing for Software Defined Infra End-to-end Closed-
Virtualized RAN Enhanced Experience with E-2-E Telco Cloud Loop Automation
Common Horizontal Telco Cloud is Now Proven
Rakuten Proof Point of scalable, reliable, flexible Telco Stack
Rakuten’s OSS and BSS
API
Cisco NSO End-to-End Service Orchestration
API API API API

GC Edge RDC/MEC WAN CDC
Cisco NSO + ESC Orchestration Orchestration Orchestration Orchestration
API API API API
Cisco VIM CVIM CVIM CVIM
RIU eCPRI WAN / MBH WAN / MBH

CPRI
Pre-Agg (GC) Agg (RDC) Central DC

~3000 ~50 2
CUPS User Plane, Control Plane, Policy,

vRAN
MEC Apps IMS, OSS/BSS
Rakuten Cloud Platform Enabled by CVIM, ESC & NSO
3rd Party Servers, Storage + 180+ VNFs from many vendors (mostly 3rd Party)
Software
Horizontal - OSS/BSS (Workflow Engine)

Zero touch, automated, fully virtualized cloud native network to offer global services
Common Multi-domain Orchestration (Cisco NSO)
RMI Network Automation

VNF Manager (Cisco ESC) Nokia CBAM
NFV Orchestration
Macro vCU SAEGW-C SAEGW-U CMS CAS TAS W-SG MRF CGNAT TAS ENUM
Radio
vDU MME ePDG VNFM SDNS WRG PNS CRDL PCRF CSCF SMS
Smallcell+
WiFi
Radio vEMS SaMOG Ultra-EM CSCF RMS
M-
STORE SDC PRS HSS SBC VxMail
SSL
Vision vDirect Load Cassandra CRDL DRA MRF CDR
Offload Cassandra
Balancer Admin Admin
(WAF)
Common Virtualization Platform (Cisco VIM)

Physical
Network
Hardware IP/MBH Routers DC Fabric Compute Orchestration
Cisco Nokia NEC Quanta Altiostar Mavenir Radware
Management
Cisco VIM Deployment Models Controller

Controller
Controller
Controller
Storage
Compute
Compute
Storage Compute
Compute
Controller
Controller Storage Compute
Compute
Compute
Compute
Edge POD
Micro POD
Centralized Infrastructure Management

Zero Touch Automated Deployment, Lifecycle Management, Software Distribution Management, Monitoring & Assurance
Central Storage
& Management
CVIM
(roadmap*)
Public Cloud
Provider
Pre-Agg. CO / Agg. / HE Regional DC Central DC Peering/Co-Lo
NanoPOD* EdgePOD MicroPOD HyperConverged POD Full-on POD MicroPOD Software Only
(1 Server) (3 Servers Min) (4 Servers Min) (100+ Servers) (100+ Servers) (4 Servers Min) (over Public Cloud IaaS)
ZERO Overhead ZERO Overhead Minimal Overhead Reduced Overhead
Cisco VIM EdgePOD for vRAN
Major Innovation enable RAN Virtualization – Zero Overhead, Real-Time Platform
Central Storage
& Management
Runs workloads
Controller + Cores on Socket 0
Compute (1)
0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 |
Controller +
Compute (2)
Controller +
Compute (3) 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 |
Additional Cores on Socket 1
Compute (N)
Runs workloads
• With CVIM MicroPOD as starting point, we removed the storage service per POD and created a central storage
• Enabled major cost reduction from 10,000s of disks, and “zero footprint” for storage at far edge sites
• Now Centralizing the management node into a Management Service
• To enable “Near Zero Footprint” for management service
• Constrained the CVIM Controller Footprint on the same 1 core per socket where host OS runs
• Enabled “zero footprint” for controller at far edge sites
• Added Hard Real Time capabilities and FPGA management support for vRAN workloads
Distributed Compute Node vs. Autonomous Cloud
DCN Approach Cisco Approach
Centralized Service
Centralized Management, Orchestration
Control, Storage Network Wide Path Computation & Orchestration, Centralized
Centralized Network Wide Resource Orchestration, Resource
Infrastructure Centralized Fault/Performance and Configuration Management Management
Management
WAN
WAN / MBH WAN / MBH
Remote Remote Remote

Compute Compute Compute All routers are autonomous with their own control, data and management plane
Centralized Service Orchestration

• Control plane across the WAN – lack of Centralized Resource Management
predictability and reliability Centralized Infrastructure Management
• Large blast radius – many remote sites

• Operational challenges – Software can fail in NFVI NFVI NFVI
many different ways

• Viability of the foundation – Is OpenStack or
Kubernetes built by the community with this WAN / MBH WAN / MBH
problem statement in mind? NO

All NFVI’s are autonomous with their own control, data and management plane
Cisco VIM – Cisco Value-add and Evolution
CVIM Values
Automated Everything
• Is CVIM just a wrap around Red Hat OpenStack Platform
(OSP)? Simple
Flexible
• No – CVIM doesn’t use RH OpenStack Platform. We just take core
OpenStack libraries from Red Hat, primarily to get RH TAC support Repeatable
Secure
• Core OpenStack libraries are open source, mostly commodities
High Performant
• CVIM uses its in-house developed capabilities to deliver
deployment, integration, performance, security, end-to-end Resilient
validation, tooling and comprehensive lifecycle management Extensible
Repairable
• Would CVIM be only OpenStack for ever?
Upgradeable
• No – Most of CVIM’s platform automation, lifecycle management,
Observable
tooling, performance and security hardening are capabilities that
are not specific to OpenStack. E2E Validated
Real-Time
• CVIM is evolving into a “Unified Cloud Platform” that will provide
Container, Bare Metal and VM workloads with a consistent set of HW Accelerated
management and automation capabilities Any Workload
Cisco VIM to Unified Cloud Platform Evolution
Virtual Machine Adding Container Adding Container on Bare
(VM) based NFV in VM Support Metal, Modularize Platform
Cisco VIM Tooling & Consistent Tooling & Consistent Tooling &
Automation Automation Automation
VNFs VNFs CNFs VNFs CNFs

(Container in (Container on
(VM) (VM) (VM)
VM) Bare Metal)
Cisco Cisco
Container Cisco Container
Platform VIM Platform
Cisco VIM
Cisco VIM Common Bare Metal
Manager
Validated Hardware Validated Hardware Validated Hardware

(Servers, Switches) (Servers, Switches) (Servers, Switches)
Shipping since 2015 We are here now Our goal in coming months
From VM → VM + Container + Bare Metal

From Private Cloud → Supporting Hybrid Cloud Deployment Model
Cisco VIM/NFVI - Benefits in Telco Cloud Stack
Horizontal Platform Consistent Virtualisation Platform from Centralised to Edge deployments
Optimized Footprint Industry-leading optimised H/W footprint, no overhead on workload resource
Workload Agnostic Multi-Cloud Platform supporting VM, Containers and Bare Metal
Programmable
Programmable API’s for complete life-cycle management of H/W & S/W
Hardware/Software
Data Plane
Fine tuning to support IO Intensive Services i.e. SR-IOV, EPA, HW Accel
Performances
Real Time
Real-Time Kernel, OS/BIOS Optimisation for Time Sensitive Apps (CRAN)
Application Support
Network Services - Telco DC Feature Requirements
Provider Edge (PE) Centralized ZTP

CIN requirements FCAPS
SR PCE/MPLS/SRv6 Transport Management Autonomous Operations
Timing (Telecom Profile) Telemetry and Analytics
• 8275.1/2 Scaling to 100/1000s of nodes
10/25G/100G - Uplink
Network
Edge
10/25 G 48 Ports Rack Depth 22 to 24 inch
L2 and L3(v6) VPN Limited Power per Rack
Line Rate NEBS Compliant
Bidi Optics Switching Physical 2 to 4 racks/edge
Networking (Openstack,K8) Physical Security/Access
1-2 micro second latency
Network Services - Node Construct Options
Public Cloud
Provider
Telco Network DC Regional/Centralised DC
PE/ PE/ PE/ PE/

WAN WAN WAN WAN
PE-ToR PE-ToR
Spine Spine Spine Spine
Leaf Leaf
ToR ToR ToR ToR ToR ToR
PE – Provider Edge TOR – Top of Rack
Network Services - Evolving End-to-End Vision
Consistent Orchestration and Policy
C-VIM C-VIM C-VIM C-VIM Virtualisation stack
DCMN/SR-NFV APIC/SR-NFV APIC APIC Controller/Management
IP/MPLS/SR IP/MPLS/SR Network

L2 ACI RL(VXLAN) ACI (VXLAN) ACI (VXLAN) Technology/Encapsulation
Interworking Interworking Interworking Interworking

Internet
SR/SRv6 SR/SRv6 SR/SRv6
SR/MPLS SR/MPLS
SR/SRv6
Public Cloud
Provider
Cisco ACI – Application Centric Infrastructure
Network Services – ACI Scale & Orchestration
ACI Remote Leaf (RL) support, Multi-Site Orchestrator, X-Domain Orchestration
Remote leaf feature stretches ACI controller domain to 100 of edge locations
Telco DC X-Domain Orchestration (NSO)
Edge WAN Central DC
Orchestration Orchestration Orchestration
ACI Multi-Site Orch.

x10
ACI Controller
Edge VIM
Leaf Leaf Leaf
IP local breakout
Remote Leaf Remote Leaf Spine Spine
Front/Mid/Back Haul (Far) Edge DC IP Core Edge / Regional DC
• Applying centralised ACI VPN, Security & service policy across 100’s of remote sites (Remote Leaf)
• Multi-site Orchestrator (MSO) – Single Pane of Glass – Policy across multiple sites/fabrics
• NSO (CFP) Core Function Pack(s) for E2E X-Domain Orchestration
Network Services – ACI Seamless Interworking
ACI seamless MPLS EVPN/Segment Routing interwork for Network Slicing
Telco DC X-Domain Orchestration (NSO)
Edge WAN Central DC
Orchestration Orchestration Orchestration
Cloud Gaming Reliable link

Low Latency Encryption
5G UPF Guaranteed BW
ACI ACI Controller
Controller (APIC)
Set VM/container policy
VPN, security, transport SLA
Nexus Leaf
Nexus Nexus Nexus
Nexus Leaf
Nexus Nexus
SR xHaul Edge DC SR IP/MPLS core Central DC

Remote Leaf ACI Fabric
• Simple, automated, scalable and open inter-domain connectivity

• Leverage ACI policies to map VNF traffic to SR traffic engineering for slicing
Network Services – SP EMEAR (Tier 1) Deployment
Deployed ACI Architecture at 34 sites Leverage Existing Capability
Remote Leaf site Multi-Site Extension of Cisco ACI to non-Cisco vGW site
• 18 Full POD Sites • Save 64x100G ports for each new vGW site.
• 16 Edge GW Micropod Sites
• Centralized policy for control • Capex & Footprint savings of 2 Spines & Controller
• Saving on Spine and Controller
plane locations • Simplified Operations with single Fabric architecture.
IP IP
WAN WAN
Extn of
Cisco Remote
leaf Solution
Remote Location Micropod On Premise Centralized location Private & Public Cloud
Security Everywhere Analytics Everywhere Policy Everywhere
Network – ACI Benefits in Telco Cloud Stack
Multi Location Deployments from Central DC Fabric to Edge Location (Remote
Deployments Leaf)
SR Integration Seamless integration with Segment Routing in the WAN
Workload Agnostic ACI integration with Openstack, Kubernetes, Bare Metal & SRIOV
Service chaining Industry leading service chaining inc. distributed Edge locations
VNF Attachment L3 fabric with on-demand expansion, GTP aware Load Balancing
Automation
Analytics & ACI integration with CVIM and NSO, ACI Telemetry & Assurance
Assurance
Network Services - Evolving End-to-End Vision
Consistent Orchestration and Policy

C-VIM C-VIM C-VIM C-VIM Virtualisation stack
DCMN/SR-NFV APIC/SR-NFV APIC APIC Controller/Management
IP/MPLS/SR IP/MPLS/SR Network

L2 ACI RL(VXLAN) ACI (VXLAN) ACI (VXLAN) Technology/Encapsulation
Interworking Interworking Interworking Interworking

Internet
SR/SRv6 SR/SRv6 SR/SRv6
SR/MPLS SR/MPLS
SR/SRv6
Public Cloud
Provider
Network Services - SR/MPLS to existing TOR/Host
MP-BGP RR hierarchy
SRv6
Server Server
ToR ToR Leaf Leaf DCIs/WAN PE
Spine Spine
Server Server ToR
ToR ToR
Server Server Leaf Leaf
Server Server Server
Server Server Server Existing DC
ToR ToR
Provider Edge function Server

Server
Provider function Server
MP-BGP RR hierarchy MP-BGP RR hierarchy
L2/L3 VPN
ToR ToR ToR ToR
L2 VLAN
OR VPP
(native or L3 OTT)
VPP SR-IOV SR-IOV
Today OR
VM vNF VM vNF VM vNF VM vNF
Longer term – technology preview

Network Services - SP EMEAR (Tier 1) Deployment
SR/MPLS/EVPN
Core Network
Service L2 Point-To-Point
• Ethernet Pseudo-Wire
• EVPN VPWS
Service L2 LAN
• EVPN
Service L3VPN
Service Multicast
PE PE
Network Services - Edge Ready Compute H/W
FCS
1HCY
2020
600mm ETSI Rack NEBS Compliant
Central Management 6x PCIE NIC
Cisco UCS C240 SD GPU & FPGA Full NVME storage
Cisco UCS C240 “Short Depth”

• Fit for Telco Cloud environmental requirements (DC, NEBS, Depth)
• No compromise on IO Capabilities: Fast NIC, NVME for CDN & Analytics
• vRAN Ready – designed to accommodate GPU & FPGA
Orchestration & Automation for SP Environments
Operations
Day 0, Day 2, Day n
Cisco Orchestration Values NFVO – Cisco Network Services Orchestrator (NSO)
Model Driven Orchestration (Model Driven Multi-Vendor Orchestrator)
Open & diverse set of interfaces
Rich multi-vendor support
NFV & Resource Orchestration VNFM – Cisco Elastic Services Controller
Security and Assurance

Flexible & Scalable (Generic VNF-Manager)
Any Virtualized Network Function

(Cisco or 3rd Party)
Cisco Virtualized Infrastructure Manager (CVIM)

(Cloud Orchestration (Openstack) with Virtual Compute, Storage & Network)
Compute and Storage Networking

(Cisco UCS or 3rd Party) (Cisco ACI, Nexus 9000, NCS5500)
Operation Model Deliverables
For both Infrastructure and Service Lifecycle Management
Mostly Operations Focused
Day 0 Day 1 Day 2-N

• Infra/Site/POD Onboarding • Network Provisioning • Auto Healing
• VNF Onboarding • DC Provisioning • Auto Healing with Restoration
• VNF Instantiation • Cloud Provisioning • App/VNF Patching
• VNF and Service Provisioning • Minor Cloud / infra update
• Licensing • Major Cloud / infra upgrade
• Service Activation • Composite upgrade (infra + VNF)
• Auto Scale out
• Dynamic Re-Provisioning
• Security management
• Licensing management
• Rollback (for many things)
……
Operations Model - Vodafone
72 hour from rack-stack-power to 1st call; In-service upgrade with 100’s of Gbps of live traffic;
VNF auto-healing; Cisco VNF + 3rd Party + IT VMs on the same cloud
MGMT SW Configuration and Cabling Validation 8 Hours
UCS Servers ACI PE/ASR9K 16 Hours
BUILD NODE (CVIM) 6 Hours
CLOUD LAUNCH (NFVI) 16 Hours
vSPGW vPCRF ASAv (FW) Gi_vDNS vCGNAT 10 Hours
INTEGRATION & SANITY CHECK 16 Hours
Total Time
FIRST CALL 72 Hours
https://newsroom.cisco.com/press-release-content?type=webcontent&articleId=1977162
Operations Model - Day 0 Lifecycle Management
Not only a CVIM POD, but Zero Touch Provisioning of an entire Edge DC
Outcome – 250+ edge data centers got deployed in ~20 days, 20 Edge DC on a single day was achieved!
Zero Touch Auto Provisioning of Cell site once RIU plugged in
NFVI NFVI NFVI
RIU eCPRI WAN / MBH WAN / MBH

1 CPRI
4 Far Edge DC Edge DC Central DC

Up!
CVIM ESC NSOFarEdge NSOE2E OSS EMS
RIU 7 vDU vCU 7
(1)
(2)
(3)
(4)
(5)
(6)
(8)
(9)
(10)
(11)
(12)
(13)
Outcome – Cell Site Secured Zero Touch Auto Provisioning – In Minutes instead of Days!
RIU – Radio Interface Unit BRKSPG-2402 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 75
Composite Major Upgrade (vRAN + CVIM) with no/minimal Impact to Radio Access
Component Time Taken / Impact Mitigation Strategy and Circular Workload
Improvement
Movement Algorithm
Terminate eNB VNF at the GC “Make Before Break”
3 to 4 min per eNB
(Service Impacting) Pre-instantiate new (upgraded SW)
eNB Before existing one is
terminated
Instantiate eNB VNF at the GC Termination and instantiation time
12-15 min per eNB.
(Service Impacting) is now out of the equation!
“Circular Workload Movement”

Update changing Kernel, Firmware,
~10 min / server (server reboot) Using the backup server.
BIOS that will require Server
+ ~ 30 min for BIOS upgrade Server reboot time and BIOS,
reboot
+ ~ 45-60 min for FPGA upgrade FPGA upgrade gets out of the
(Service Impacting)
equation!
“FH Control Channel enhancement

to avoid RIU reboot”
RIU reboot and establish RIU vDU signals the RIU to hold on
connectivity with the vDU ~10 min. during connection flip over during
(Service Impacting) make before break.
RIU reboot time and GPS sync time
gets out of the equation.
vDU, vCU (and EMS) connectivity “Make Before Break”

establishment to make sector fully ~1 min. vCU/vDU/EMS interconnection
operational (Service Impacting) time gets out of the equation
Outcome – 2 hours+ outage per cell sector to be reduced to <2-3 minute

Cisco Overall Telco Cloud Stack Benefits
Cloud Infrastructure Network Services Operations Model
Pre-Integrated Faster TTM & reduced risk in deployment (enhanced with CX Bluestack)
Open Independent NFVI components, leverage open interfaces, 3rd Party VNFs
Best of Breed Every component is best of breed in its category (NSO, CVIM, UCS & ACI)
Centralized management and Lean Edge supports distributed Edge &

5G Ready MEC
Evolved
Evolved planning/design, change management, problem resolution
Operations
Key Takeaways
Key Takeaways
• Service Edge Transformation is under way with leading operators driving
• Need more focus with multiple Industry Initiatives, forums and technologies
• 5G is serving as a catalyst in the market (not all use-cases are 5G related)
• Service Edge requirements are changing but infra. use-cases still prominent
• Cisco has Industry-leading Platform based on real-world deployments
• Cisco is continuing to evolve to meet changing requirements with use-cases
like CRAN, Cloud CMTS, CUPS based gateways, Gaming, vCDN, AI/VR
Far-Edge Edge Core Cloud
Complete your
online session
survey • Please complete your session survey
after each session. Your feedback
is very important.
• Complete a minimum of 4 session
surveys and the Overall Conference
survey (starting on Thursday) to
receive your Cisco Live t-shirt.
• All surveys can be taken in the Cisco Events
Mobile App or by logging in to the Content
Catalog on ciscolive.com/emea.
Cisco Live sessions will be available for viewing on

demand after the event at ciscolive.com.
Continue your education
Demos in the
Walk-In Labs
Cisco Showcase
Meet the Engineer

Related sessions
1:1 meetings
BRKSPM-2743
Opening Keynote 09:00 5G Cloud Native
Packet Core and 8:30
11:00 BRKSPM-2012 Network Slicing
5G xHaul Transport
11:00
BRKSPG-2402
The SP Service Edge BRKSPM-3295 11:15
Transformation with 5G, 5G Timing & Synchronization
virtualization
BRKSPM-2539
8:30 architectures 09:00
Mobile Virtual Packet Core -
LTRSPG-1226 Deployment Challenges and
TECSPM-2306 BRKSPG-2518 NFV Management and PSOSPG-2188 Best Practices on Openstack
5G System – Cisco SP Programmable 14:00 Orchestration (MANO) Addressing the
/ VMware
proposal and SDN solution for the with Cisco NSO, ESC Architectural 13:30
technology in every Metro Fabric 14:30 Transitions within
domain Service Provider BRKSPG-1565
BRKSPM-2010 Networks. A Deeper Dive into the 11:30
14:45 Applying Security in a Telco Cloud Architecture
BRKSPM-2786 5G World 17:00 Evolution to support 5G
Unleashing the Power 17:00 and MEC
Guest Keynote
of a Software-
Defined 5G Network Cisco Live
Architecture Celebration 18:30 BRKSPM-2007 11:30
Cisco IOT in a 5G World
5G network
transformation
© 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 82
Thank you

BRKSPG 2402

Uploaded by

Document Informationclick to expand document informationnfv9

Document Informationclick to expand document information

Copyright:

Available Formats

BRKSPG 2402

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BRKSPG 2402

Uploaded by

Copyright:

Available Formats

The SP Service Edge

Brian Meaney Distinguished Systems Architect

Centralized DCs / Clouds

SP Multi-access Edge IoT Edge Enterprise Edge Cloud Services Edge

Edge on the Devices

• Market Trends, Analysts Viewpoints

Explosive growth Advent of 5G Changing Rise of cloud Cyber security

Low Latency for better QOE and to

• Open vRAN ecosystem - accelerate adoption of

Distributed Connectivity Control

Worldwide Container Instances Installed Base by Virtualized/Non-

Traditional operations staff will not be able to

Traditional Infra IaaS/PaaS/SaaS

Private Cloud Private Cloud

Mgmt. and Ops Public Cloud 2

Industry Reference Project Reference

1 Do it Yourself (DIY) 2 Vertical Solution Stack 3 Common Horizontal Platform

MANO MANO MANO MANO Common MANO

Virtualization Virtualization Virtualization Virtualization Common VIM / Virtualization

Hardware Hardware Hardware Hardware Hardware Hardware Hardware

Majority of the market today belongs Here

SP’s want to get Here

NFV and SDN DC Architecture Cloud Agility

CO - Central Office CORD – Central Office Re-architected as a DC

Latency Reduction Data Reduction Offload at Edge

~ 1 – 30 ms ~ 100 – 200 ms Anchor Server

App App Infra workloads UPF

Enterprise Location Premises

NFV and SDN DC Architecture & Cloud Agility

Edge Computing Cell Site CRAN Hub/ CO / Agg. / Regional DC Central DC

MEC is an architecture moving services closer to the user to the vDU

EMEAR Frontend Data Backend Data Center

Distributed Telco Cloud

Slicing, Flexible deployment,

Push data plane to the edge,

CUPS and MEC Network Slicing Cloud RAN

What? Distributed workloads for Logical, virtually Functional split and

Why? • Services (Low latency, QoE) • Services, Monetization • Lower costs

Network • Any-to-Any connectivity • End2End SLAs, Steering • Low latency

CUPS: Control/User Plane Separation

• Decompose the Traditional gateways

user planes. IMS

• Control and User Plane Separation Access Access

• Enable User Planes (UP)s to be

• Augment UPs with with in-line

disaggregation are enabled with Access Access

EPC/5GC EPC EPC/5GC

To Mobile Core To Mobile Core

• Operator Lead – Drive Open, Standard Interfaces

TIP • Operator Lead – Intel providing major input

• Cisco Lead, Operator Driven – Compliment ORAN/XRAN

Wireline Slice Common

• New service introductions are quicker

UPF Svcs Data

Cross-Domain Automation & Orchestration

AR/VR/ mMTC Slice

Access Aggregation Core