Day_01_6_lamoureux_b.. - Upperside Conferences

WHEN THE CLOUD
REACHES THE RAN
IMPACT OF CRAN ON CLOUD-BASED ARCHITECTURES
Pierre Bichon
Philip Lamoureux
PARIS - MAY 21ST, 2014
THE CONTEXT OF CLOUD RAN
3 trends are currently happening in the IT industry:
1) Cloud Computing
2) Cloud Networking
3) Cloud Access
Only when all 3 join and leverage each other, can we reach 5G
DATA CENTER VIRTUALIZATION
Image
Repository
Dashboard
Network Node
Registry
DHCP / IPAM
Provides UI for
Services & Modules
Compute Node
Hypervisor
VM
Router / GW
Block Storage
Scheduler
Block
Storage
DB
Scheduler
Conductor
Cell
Image DB
Queue
Object Storage
Compute
DB
Proxy Server
Queue
Volume
Orchestration
Provides Authentication
And service catalog
for All Services & Modules
Identity
Object
Store
Metering
DATA CENTER NETWORKING
DC2
DC1
MOBILITY OF
RESOURCES
DC3
Cloud
Networking
LOAD
OPTIMIZATION
Openstack Typical Deployment Topology
EXTENDING THE DC WITH NFV
Data Center Orchestration
Network Automation
VM
VM
VM
Data Center
Network
VNF
VNF
VNF
Network Virtualization
VNF
VNF
VNF
Data Center
Interconnect
VPN
VPN
VPN
Service Virtualization
Network “Edge”
DC Edge
Access / Aggregation / Core Network
Users (Humans, Things)
SERVICE VIRTUALIZATION & CHAINING
NFV use cases*:
Virtual Application as a Service
Virtual Infrastructure as a Service
Virtual Network Platform as a Service
vEPC, vCPE, vContent, vFixed Access
Virtual Network Function as a Service
Virtual Network Function Forwarding
Graphs (Service Chaining)
Other use cases:
Security
Subscriber Management
Custom services developed by Operator
Centralize or distribute services for maximum flexibility
Monetize services by enhanced QoE
Orchestrator
Distributed
Services
Svc1
Centralized
Services
SDN Contr
Svc2
Svc3
PCRF
Mobile Core
Services Complex
P-GW
MME
CSR
RAN
SGi
AggRouter
S-GW
Virtualized
* ETSI GS NFV 001
Pre-provisioned
Dynamically
Adjusted
Horizontally
Scaled
macro
Intra/Inter
Carrier Data
Centers
SON
CRAN, DAS
THINGS
HUMANS
FOUNDATIONAL: A SEAMLESS, INTELLIGENT NETWORK
Internet
indoor / outdoor
small cells
Enterprise
public / private
small cells
user access layer
Coordinated prediction
VIRTUALIZATION OF THE SERVICE ANCHOR
network access layer
access
routers
service anchors
long-haul
microwave
short-haul
microwave
LOS, NLOS,
P2P, P2MP
7
service anchors
transport nodes
Third party
access
service anchors
Seamless
MPLS
transport nodes
Access & Aggregation
Core
SON
SON
AUTOMATION, PATH COMPUTATION, VISIBILITY
Copyright © 2013 Juniper Networks, Inc.
www.juniper.net
Service
Virtualization
Layer
WHEN THE CLOUD REACHES THE RAN
Virtualized Access Network
Virtualized Core
HetNet
Distributed Data Centers + Interconnect
Macro
Service
HUMANS
Smart Phone
Service
Tablets
Service
Clusters
?
Mobile
Router
Mobile
Base Station
Internet
Service
Service
SC
4G Small Cell/
WiFi
Enterprise
Service
Instrumentation/M2M
Service
THINGS
Intra/Inter
Carrier Data
Centers
Connected Enterprise
WiFi/Femto
Enterprise/SMB
Residential
Architectural Layer Evolution
Simplification of layers
Connected Home
Simplification of protocols/interfaces
WiFi/Femto
Flexible architecture based on software
Enterprise
Residential
Branch
CONVERGED ACCESS NETWORK
Characteristics of the converged user access network
Macro Cells
(2G, 3G, LTE)
Baseband
Pooling
(C-RAN)
2km+
100s-1000s
subs
RRU
CPRI
BBU
Micro/Pico Cells
Indoor
Out-door
<500m
100s subs
Public Wifi
Indoor
Out-door
<100m
10s subs
Private Wifi
Private Pico cells
<100m
10s subs
Femto Cell
WIFI
<20m
<10 subs
Commercial
And Residential
Broadband
tethered
1-1000’s subs
Trend towards heterogeneous networks and
seamless user experience
User experience:
Best coverage and highest rates at every user location .
Presence & application awareness –context aware
Seamless handover
Network Implications:
Overlapping Macro/Micro/Pico/Relay/Femto cells and Wifi
Services
Drastic increased application usage and throughput per
user
Fast, seamless handoffs between technologies (iRAT)
and between Macro/Micro/Pico/Relay/Femto.
Handoff between private and public networks
Potentially large increase in cell count arising from small
cells
Serious Performance, operations, and management
challenges for conventional hierarchical networks
C-RAN FRONTHAUL AND BACKHAUL
Challenges for Transport
RAN Transport
1588v2
Grand Master
Aggregation of
Traditional RAN
20’s of sites served by a BBP
- 10k’s of subscribers.
Baseband pool (BBP)
Hub site
MicroCRAN
Backhaul
10/40/100G
DWDM
Fronthaul: Very Low Latency,
High Capacity - direct link
CPRI/OBRI
WDM
Fronthaul
CPRI/OBRI
RRU
Small cell
RRU
Macro CRAN
RRUs
Backhaul: Controlled
Latency, High Capacity, High
Availability
Fiber
MMW radio
Coordinated /centralized
timing
Support “traditional” RAN
CRAN SHARING CONSIDERATIONS
Opportunities to Architect a Share-able infrastructure
RAN Sharing
RRH
Towers
Shared BBU Pool
(and CRANaaS)
BBU
Shared DWDM
C
P
R
I
NFV
Today: Tower, Backhaul
and RAN sharing
CRAN future: Tower, RRH,
fiber infrastructure, BBU
Pool, Cloud
•MNOs determine
what NFs are located
at Hub sites
C-RAN USE CASES
Macro scale deployment will require use of SDN based automation and
virtualization tools
Source: Heavy Reading, C-RAN and LTE-Advanced: The Road to True 4G and Beyond, Oct 2013
C-RAN BACKHAUL WITH PCE
Optimizing the connectivity between fluid baseband pools
PCE based Network SON
SON
PCE
Analytics
Each BBP will cover a large
geographic area ~20 macro
sites
Baseband pool
(BBP)
RRU
CPRI/OBRI
CPRI/OBRI
RRU
BBPs will be connected
through high capacity
backhaul links
To maximize the utilization of
scarce spectrum the BBP
capacity will be dynamically
updated
SDN based PCE will manage
the backhaul links between
BBPs as SON is managing
radio resources (RRU)
PCE: Path Computation Engine
NODE MANAGEMENT AND AUTOMATION
Before: Micro-Management
Features and Benefits
NMS
• Open standards (NetConf/Yang) based
architecture
• Manage 1000’s of devices through one
Junos powered device
• Reduce touch points, protocols, time,
and “CLI languages”
Now: Simplified
NMS /
SDN
controller
Fusion
Technology
• Converge network layers for reducing
the complexity and cost of operations
• Increase responsiveness to dynamic
events in the network
• Improve reliability
• Manage Junos and 3rd party platforms
(IP, optical, mobile, cable etc)
MANAGING FRONTHAUL WITH VNEM
OAM complexity can be drastically reduced through vNEM approach
SON
vNEM
PCE
vNEM Converges Layers
Config
Analytics
database
Each BBP will connect to
100s of RRU through CPRI
(fronthaul) links
Baseband pool
(BBP)
RRU
CPRI
CPRI
RRU
Within each BBP region
there will be a large number
of microwave and macro
base stations
Individual OAM of these
large number of devices will
be complex and costly
Open standards (NetConf/
Yang) based vNEM can
significantly reduce the
complexity and cost
vNEM: Virtualized Network Element Manager
CONCLUSION
1) Cloud Computing
2) Cloud Networking
Same concepts
Rely on each other
3) Cloud Access
Cloud Networking and Cloud Computing are building
fundamental technologies that will make Cloud Access
successful and move to the next step, beyond a
centralized BBU