Towards vRAN / Edge / 5G vRAN... · data processing compute nova networking cinder glance swift neutron ironic storage block image object bare-metal provisioning horizon tripleo dashboard

Telco Technology Office Techies:François DuthilleulFred HerrmannTimo JokiahoIgnacio Verona

Towards vRAN / Edge / 5GRed Hat EMEA Telco Vertical

February 2019

ZA Telco Account Team Edgar KamupiraGavin McDougall

Presenter

Presentation Notes

telco icons: https://drive.google.com/drive/folders/172g3ITUkKFAsHMhDHA5m875FLJR9IwXl?usp=sharing

#TTO #vRAN #Edge #5G2

● Introduction ● Red Hat Development Model >● Red Hat Position in Telco Networking Industry, Network Function Virtualization (NFV) >● Highlights of The Work Done in RHOSP During NFV Journey >● Overview of vRAN models >● Red Hat @ The Edge >● 5G, High Level >● Some Technical Requirements Spanning over vRAN / Edge / 5G >● Seamless Support for Virtual Machines and Containers >● Red Hat Openstack Platform @ The Edge / NFV >● Red Hat OpenShift Platform @ The Edge / NFV >● Red Hat participation in vRAN / Edge / 5G Industry Work >● Red Hat Telco Customers & Ecosystem >

Contents


EMEA Telco Technology Office (TTO)

Telecom ExpertiseRAN, EPC, IMS, 4G, 5G, CPE,...

NFV Transformation

ETSI NFV/MEC,LF Networking, O-RAN, TIP, ...

NFV InfraSR-IOV, OVS-DPDK, EPA, RT-KVM, Security, Multisite, Upgrade, Service Assurance,...

TTO


Much of this is enabled by Open Source

Software Defined Networking (SDN)

Network Functions Virtualization

5G networksEdge / vRAN

IoT, M-to-M, Mobile Applications

IPTV, DVR & intelligent Content Delivery Networks (vCDNs)

Telecom Industry is Investing

Red Hat Development Model


PARTICIPATEINTEGRATE(community platforms) (supported products

platforms, and solutions)

INTEGRATE(community platforms)

STABILIZE(supported productsplatforms, and solutions)

STABILIZE

From Community to Enterprise: Upstream FirstRed Hat Development Model


Red Hat model is staying on mainline

Branch and rebase

and rebase and rebase ...

Mainline and Forking

Branching Strategies


Red Hat Open Advantage

38,968 or 17%

2,052 or <1%

24,662 or 11%

OpenStack — Contribution Otaca Release

www.stackalytics.com

79,455 or 66%

35 <1%

1,335 or <1%

Ceph— Contribution Since Inception

www.metrics.ceph.com

1,639 or 12.1%

<100 or <1%

Linux Kernel 4.3 — Contribution by changesets

https://lwn.net/Articles/654633<100 or <1%

169 or 24%

<25 or <1%

KVM— Contribution by Employer

KVM Forum 2015– Keynote (Paolo Bonzini) <25 or <1%

4,330 or 14%

66 or <1%

209 or <1%

Kubernetes — Contribution Since Inception


1,214 or 4%

8 or <0.1%

21 or <0.1%

Docker — Contribution Since Inception


516 or 4%

12 or <1%

0

OVS (Open vSwitch) — Contribution Since Inception


798 or 9%

382 or 4%

182 or 1%

OPNFV — Contribution Since Inception


(RedHat + Ansible) – 8,242 or 32%

0

10 or <1%

Ansible — Contribution Since Inception


2,384 or 7%

OpenDaylight (ODL) — Contribution Since Inception

http://spectrometer.opendaylight.org<100 or <1%

<100 or <1%

Red Hat Posit ion in Telecom Networking Industry, NFV

#TTO #vRAN #Edge #5G

Networkscore | edge

Internal IT, OSS/ BSS

SE

CU

RIT

Y -A

UTO

MA

TIO

N -

CLO

UD

NA

TIV

E

VNF

Cloud Services

vApp

cAppCNF cApp

PNF

MANO

integrat ion

applicat ions

platforms

management

consumer

B2Bcustomer

Developer

Operat ions

Gov IoT Video Partner/ 3rd Party Services/ SaaS SE Mobile

Apps Web

SDN

Red Hat Portfolio @ Telco

10

Presenter

Presentation Notes


● Revolutionizing the traditional building practice of Network Equipment and enabling the interchangeability / freedom of choice of Network Functions

● Building the path towards 5G innovative Network Services by enhancing QoS, Bandwidth & Latency for IoT, M2M, Content Delivery (e.g. VoD)

Business Motivation for NFVNetwork Functions V irtualization (NFV) is a network architecture concept that proposes using IT virtualization related technologies, to virtualize entire classes of network node functions into building blocks that may be connected, or chained, together to create communication services.


RH Focus:Trusted BaseSW Platform Provider

Red Hat works with partners

Red Hat works with partners

KVMHypervisor

Red Hat Enterprise Linux as Host

RHELGuest

RHELGuest

Red HatStorage

DPDKOpen vSwitch

ODL ML2libvirt

JBossOpenShift(tooling)

Open & AgileOpen, Flexible, Modular,

Programmable

Single Platform for multiple use cases beyond NFV

Carrier Class Architecture

Performance, Throughput, Scale, Optimised for NFV

High Availability & Security

12

Red Hat on ETSI NFV ISG Reference Diagram

OtherGuest

#TTO #vRAN #Edge #5G Cert ified Red Hat OpenStack Platform plugins: https:/ / access.redhat.com/ art icles/ 1535373

IaaS+

IaaS

TELEMETRY ORCHESTRATION

TELEMETRY SAHARA HEAT

DATAPROCESSING

COMPUTE

NOVA

NETWORKING

NEUTRON IRONICCINDER GLANCE SWIFT

STORAGE

BLOCK IMAGE OBJECT

BARE-METALPROVISIONING

HORIZON TRIPLEO

DASHBOARD

SHARED SERVICES

IDENTITY

KEYSTONE

DIRECTOR

DEPLOYMENTand

MANAGEMENT

MANILA

SHARED FILESYSTEM SECRETS

BARBICAN

LOAD BALANCING

OCTAVIA

QEMU-KVM Libvirt DPDK

R ed Hat E nterprise Linux

Red Hat Openstack Platform (RHEL Always as a Host)

13

https://access.redhat.com/articles/1535373


Windows Windows WindowsLinuxLinux

SUPPORTED GUESTS

OpenStack

RHEL + KVM Ceph OVS

Storage Network

SERVERS

Virtualization Security Ecosystem Network Storage

KVM Network Stack

Device Drivers

LINUX KERNEL

Security Enhanced Linux (SELinux)

A typical OpenStack cloud is made up of at least 10 coreservices + plugins to interact with 3rd party systems

Co-engineered with RHEL


Red Hat is the only company which is one of the leading contributor on all upstream projects, which constitute a consistent, high quality, high performing

NFVI / VIM platform based on Openstack.

Openstack Platform is a collection of about 10 Open Source projects, all having their own release schedules, governance models etc. To be able to put together a consistent Openstack Platform, you

need to understand all of these projects in detail not just Openstack

Integrating it All Together

Work Done for OpenStack During NFV Journey


PERFORMANCE AVAILABILITY SECURITY MANAGEABILITY LIFECYCLE

Enhanced Platform Awareness (EPA) SR-

IOV, OVS/DPDK, vCPU, NUMA pinning, Huge

pages, RT-KVM ...

High AvailabilityFault Tolerant Design

Enterprise Hardened CodeInstance Availability

End-to-EndSELinux sVirt

Neutron Security GroupsBlock EncryptionSSL/TLS on APIs

LoggingPerformance Monitoring

Operational Visibility Policy and Compliance

OpenStack LifecycleUpdates/Patches

--> Absolutely Relevant and Necessary in vRAN / Edge / 5G Space as well

Now we are doing the same things to make OpenShift NFV Ready !!!

To Make OPENSTACK NFV Ready, Development During Last Years


virtio0

DPDK kernel

OVS-DPDK

OpenStack APIs

compute node regular NICs

virtio1 VF1DPDKkernel

VF0

regular NICs

DHCP+PXE

Base (VNFs management: ssh,

SNMP, logs)

eth0 eth0

bondedbonded

bonded

VNFc0 VNFc1

fabric0: user traffic(provider network)

fabric1: user traffic(provider network)

bonddataplanedataplane

PF0 PF1

SR-IOV Host/VNFs guests resources partitioning

NFV Deployments Today: OVS-DPDK and/or SR-IOV


GLOBAL COMPLIANCE DRIVEN RESOLUTION OF SECURITY REQUIREMENTS

Compliance Driven Approach• The global nature of our OSP

deployment places a diverse set of security requirements on our product.

• Our approach here has been to analyze the top compliance init iat ives globally and then build a plan to address them in a cohesive and focused manner.

Compliance Driven Approach

Technical Security Controls Resolution

Solving for the “Big Three”

Focus is on: FedRAMP, ANSSI and ETSI• The technical control groups are similar

across the big three.

• Solving for these should get us 90% of everyone else.Top Technical Control Groups

Addressed• Infrastructure and Virtualization S ecurity• E ncryption and K ey Management• Identity and Access Management• Threat and Vulnerability Management

OpenStack Security Strategy


● Ansible is a simple yet powerful and

agentless automation framework.

● Also for NFV! Modules available for

multiple VNFs

● Easy integration via API to current

tooling and processes

Enabling DevOps and GitOps through the

complete organization (infra, QA, AppDev)Specific features for

Network Device

Management, like:

● SSH CLI

● Netconf

● IPAM plugins

Automation IS A Critical Part of NFV


Every 6 months we release a version of Red Hat OpenStack Plat form supported for 1 year

E very 18 months we produce a “long life” version, which customers can opt to have support for up to 5 years

Long Life Long Life Long Life

RHOSP 8Liberty

RHOSP 9Mitaka

RHOSP 10Newton

RHOSP 11Ocata

RHOSP 12Pike

RHOSP 13Queens

RHOSP 14Rocky

RHOSP 15Stein

RHOSP 16Train

3 years 3 years 3 years(+2 years) 1 year 1 year 3 years

(+2 years)* 1 year 1 year 3 years(+2 years)*

RELEASEDJune 27,

2018

RELEASEDNovember 13,

2018

ON DEV

21

RHOSP Lifecycle And Support

Presenter

Presentation Notes

Radio Access Network (RAN) CLOUDIFICATION

#TTO #vRAN #Edge #5G23 NOTE: No voice circuits or related elements shown, only packet data related

2/3G RAN

4G RAN

Digital Mobile Networks Evolution, 2/3G+4G


Distributed RAN

http://xhaul.eu/wp-content/uploads/2015/10/workshop-FSAN-09-2015-atlanta-orange.pdf

Full base station functionality distributed on each antenna sites -> expensive, no scalability, radio interference control challenges, etc ...

BaseBand Units @ Antenna Site BaseBand Units @ Central Office (Cloud)

Part of base station functionality (BaseBand Units, BBU) centralized (base station split).

X2: “fast” handover protocol between base stationsD-RoF: Digital Radio over Fibre

Target for NFV, virtualization or

containers

High Level Overview On vRAN (4G / LTE)

Centralized RAN

24

Presenter

Presentation Notes

Slides from Orange: - http://xhaul.eu/wp-content/uploads/2015/10/workshop-FSAN-09-2015-atlanta-orange.pdf

http://xhaul.eu/wp-content/uploads/2015/10/workshop-FSAN-09-2015-atlanta-orange.pdf


Virtual Machinesor

Containers

vBBU

Industry is driving to standardize this interface to have freedom to acquire RRH and vBBU from different vendors(dis-aggregation)

CoreNetwork

S1 Interface

High Level View to Dis-Aggregation (LTE/4G)

Target for NFV (virtualization or containers), consistent with Core

Network NFV deployment

25


ALTIOSTAR Presentation @ Telenor, Oslo June 8th, 2016Cloud-RAN (vRAN) and Application Intelligence (AI)

ETSI PoC #34 - SDN Enabled Virtual EPC Gateway


https://www.itu.int/dms_pub/itu-t/opb/tut/T-TUT-HOME-2018-PDF-E.pdf

Mapping of CU and DU functions according to the split points:- 5G(a): high layer split (F1)- 5G(b): low layer split (FX)- 5G(c): cascaded split

Option 2(3GPP)

6 7 8

ITU-T: Evolving From 4G To Split Function In 5G

https://www.itu.int/dms_pub/itu-t/opb/tut/T-TUT-HOME-2018-PDF-E.pdf


CU ● Distributed Cloud, Low Footprint

DU ● Realtime, PTP, HW Acceleration (FPGA, Smart NICs)

RAN Requirements on Telco Cloud

+

RAN Cloudification ● vBBU (4G), vCU/vDU (5G)


Why does Red Hat Care ?• Enables virtualization of the BBUs → NFV implementations

of pooled functions• S plit strategies have significant impact on key system

aspects that may affect NFV infrastructure, such as:• Latency Const raints, including real-time

requirements (R T kernels, network latency)• Synchronizat ion requirements• Throughput requirements (performance Mpps &

Gbps, host bypass mechanisms, acceleration)• NFV-PoP dist r ibut ion requirements (distribution

strategy & number of nodes per site / network tier)• Reliabil it y & Availabil it y, Service Assurance

and QoS related requirements• HW interface and HW assist requirements

vRAN, Why Should I Care?

Why do Operators Care ?• Enables CapEx savings – e.g. through site

acquisition cost and function pooling related gains as well as simpler electronics on antenna site

• Enables OpEx savings – e.g. through site maintenance, spare part logistics on E dge DC and power related gains on cell site

• Advanced interference control techniques such as eICIC (enhanced Inter-Cell Interference Coordination) will benefit from the parallelism and increased processing power at the centralized BBU

• Carrier aggregation and CoMP (Coordinated Multipoint) will leverage the operational and spectrum efficiencies of resource sharing at the centralized processing unit


https://www.redhat.com/cms/managed-files/ve-virtualized-radio-access-network-altiostar-partner-solution-brief-f11705-201803-en.pdf

ALTIOSTAR - Red Hat Partnership

https://www.redhat.com/cms/managed-files/ve-virtualized-radio-access-network-altiostar-partner-solution-brief-f11705-201803-en.pdf


https://multi-rat--network-terminal-function-virtualization-02.webself.net/file/si230266/download/2.%20vRAN%20ETSI%20workshop%20De%20Lannoy-fi4516387.pdf

ALTIOSTAR Trial @ Orange Poland

https://multi-rat--network-terminal-function-virtualization-02.webself.net/file/si230266/download/2.%20vRAN%20ETSI%20workshop%20De%20Lannoy-fi4516387.pdf


Intel HW

Red Hat Enterprise Linux

Red Hat OpenStack Platform

OpenShift

vSBC (P-CSCF)vS/PGW

vPCRFvBBU vMME vTAS

vI/S-CSCFISC

MwMb

SGi / Mb

Gx Rx

S1-MME

S1-U S11

VoLTE End-to-End ArchitectureMWC 2018

vRAN, VNFs on VMs, Network Function on Container (CNF)


Rakuten Cloud Native Mobile Network

Earlier Press Release:https://rakuten.today/blog/rakutens-upcoming-end-to-end-cloud-native-mobile-network.html

Great video, where Rakuten CTO Tareq Amin speaks:https://www.youtube.com/watch?v=adr8k0cVSTQ

Most recent Press Release on Feb 12th, where the partners are disclosed, including Red Hat:https://global.rakuten.com/corp/news/press/2019/0212_06.html

https://rakuten.today/blog/rakutens-upcoming-end-to-end-cloud-native-mobile-network.html

https://www.youtube.com/watch?v=adr8k0cVSTQ

https://global.rakuten.com/corp/news/press/2019/0212_06.html

34

Red Hat @ The EDGE


Telecommunications Manufacturing

Transportation Health - Life Science

+Retail, Public Sector and others

But Here We Talk About Telecom Networking

Edge Is Not Specific to Any Industry

Presenter

Presentation Notes

While Telco is currently a focus area and major revenue opportunity, we have customers from a wide range of industries, incl. industrial IoT, healthcare, the military, etc. These industries’ use cases have very different requirement sets, but none of their requirements is unique to them. Others have similar requirements. We therefore solve for all these industries with a common platform offering.


vBBU

Wifi/DAS

Ethernet or Fiber Fronthaul

Virtualized RAN: Virtualized BBU (LTE/4G), CU/DU (5G)

eNodeB

5G RAN

LTE/4G

Fiber / Coax

CU

Functional Splits

ePDG

DU

RU

RU

Fronthaul Midhaul

Backhaul

DU

LTE RAN

Regional or Core DCFor LTE EPC or NG-

Core (5G)Edge Compute

Edge Compute

Edge Use Case - vRAN for LTE and 5G

36


Form Factors

Deployment

Workloads

➢ Enable multiple deployments of Red Hat platforms at the Edge

➢ Adapt Red Hat platforms to all Edge form factors

➢ Enable Red Hat platforms to run all kinds of Edge workloads

Red Hat Edge Strategy as Software Platform Vendor

37

Presenter

Presentation Notes

NFV Strategy: 100% virtualized, 100% software, i.e. get rid of SR-IOV 100% software virtualization stack - Challenge is the performance HW dependency creates huge testing matrix for VNFs: TEF, Verizon Agility/Velocity in new features development Strategy: provide a DPDK based vSwitch providing virtio interfaces Get the door open to HW offload as long as it’s transparent/virtio-standard from VM/container perspective Go live with OVS-DPDK and migrate to VPP Improve DPDK/virtualization performances and features (benefit to any DPDK based vSwitch) Note: VM or containers have the same vSwitch needs, and we provide them virtio interfaces


Option 1: independent clustersFully Distributed OpenStack

RC#1

(RT)

RC#2

(RT)

RC#≤

NR(

RT)

Remote Server Pool

EN#1

(RT)

EN#2

(RT)

EN##

NE(

RT)

EN#3

(RT)

Central Server Pool

Fronthaul(Network)

Mid | BackhaulNetwork

Option 2: remote computeCentralized OpenStack

Remote Server Pool

EN#1

(RT)

EN#2

(RT)

EN##

NE(

RT)

EN#3

(RT)

Central Server Pool

Fronthaul(Network)


≤100us≤20km(CPRI,eCPRI)

≤2..5ms≤200km

(typ. Req)

Option 3: hybrid of 1 and 2

[SMALL] Remote Server Pool

EN#1

(RT)

EN#2

(RT)

EN##

NE(

RT)

EN#3

(RT)

Central Server Pool

Fronthaul(Network)

[LARGE] Remote Server Pool


RC#1

(RT)

RC#2

(RT)

RC#≤

NR(

RT)

RC#1

(RT)

RC#2

(RT)

RC#≤

NR(

RT)

RC#1

(RT)

RC#2

(RT)

RC#≤

NR(

RT)

OpenStack & OpenShift @ The Edge

38

RHOSPFullRHOSP

DCN

RHOSPFull

RHOSPFull

RHOSPFull

RHOSPFull

RHOSPDCN

5G, High Level


Some Fundamental Use CasesARTIFICIAL

INTELLIGENCE

VIRTUAL RAN(vRAN)

ENTERPRISE EDGE

AR / VRMEDICINE

ENHANCED MOBILE BROADBAND

MASSIVE MACHINE TO MACHINE

CONNECTED DRONES

ULTRA RELIABLELOW LATENCY COMMS

Building ultra-reliable immersive experiences for people and objects when and where it matters

5G Will Transform Our Lives

40

Presenter

Presentation Notes

Reminder – talking points we’ve been working toegher with ATT for years -- communities , joint sessions, POC’s Right tool for the right job .. Role for OSP, OCP, ONAP,… to build the whole framework for a distributed cloud native architecture 2) What we have demonstrated with openshift and non-network apps Demos / efforts on network apps Ongoing work for network apps


Enhanced Mobile Broadband

Massive Machine Type Communications

Ultra-Reliable Low Latency Communications

Differing Requirements for 5G Use Cases Enhanced Mobile Broadband:User Experienced Data Rate, Peak data rate, Spectral efficiency, Traffic density, Network energy Efficiency, MobilityMassive Machine Type Communications: Connection Density, Low Cost, Low Power ConsumptionUltra-Reliable Low Latency Communications:Latency, Reliability, Mobility

Three Fundamental Use Cases of 5G

41

Presenter

Presentation Notes

Reminder – talking points we’ve been working toegher with ATT for years -- communities , joint sessions, POC’s Right tool for the right job .. Role for OSP, OCP, ONAP,… to build the whole framework for a distributed cloud native architecture 2) What we have demonstrated with openshift and non-network apps Demos / efforts on network apps Ongoing work for network apps

#TTO #vRAN #Edge #5G © 2016 Affirmed Networks, Inc. All rights reserved.

Pre-Paid ConsumerPost-Paid Consumer

Fixed Wireless Access –Residential Fixed Wireless Access- Small Business

Consumer MVNO 1

Car ManufacturerFarm Conglomerate

Mining Company

Federal Government Devices

Smart City 1

Public Safety Video Surveillance Self Driving Cars

Mission Critical Application-Health Industrial Automation

Augmented Reality

Cloud Native Architecture allows for a low “minimum cost of entry” per slice –

Ability to cost effectively scale to thousands of slices of all

different sizes

LLC/URLLC

IoT/m-IoT

MBB/eMBB

Micro-Segmentation With Slicing

42


● A given NF may be deployed closer to the edge depending on network slice requirements ⇒Importance of homogeneous NF execution environments

● Network Slice LCM (multi-site) and isolation (QoS, Resource Allocation) requirements

5G Network Slicing - Including Radio Interface

43 https://www.ericsson.com/assets/local/publications/white-papers/wp-5g-systems.pdf

https://www.ericsson.com/assets/local/publications/white-papers/wp-5g-systems.pdf


5G User Plane Function (UPF)

China Unicom @ OpenStack Summit 2018, Berlin

Same Function at Different Locations

44

Presenter

Presentation Notes


eNB

MME

SGW PGW

HLR

HSS

OCF

PCRF

CP CP

DP DP

Box / Device centricLTE/4G

5G RAN4G

RAN Localized GW or Central GW Data Plane

Control Plane – Mobility, Sessions & Service Management

AUSF AFPCF

5G - Cloud Based

OpenStack or KVM OpenStack

vBBU vMME vSGW vPGW

vPCRFvHSS/HLR

• CP-DP Separation• UPF is controlled by

AMF and SMF• Data plane extensibility

45

Mobile Packet Core Evolution Towards 5G


API Management

Security

Cloud Native / Containers

AF Application FunctionAMF Access and Mobility Management FunctionAUSF Authentication Server FunctionDN Data NetworkFE Front EndNEF Network Exposure FunctionNRF Network Repository FunctionNSSF Network Slice Selection FunctionPCF Policy Control Function(R)AN (Radio) Access NetworkSEPP Security Edge Protection ProxySMF Session Management FunctionUDM Unified Data RepositoryUDR Unstructured Data Storage FunctionUE User EquipmentUPF User Plane Function

46

5G - Service Based Architecture (SBA)

Some Technical Requirements


Latency

consumer ⇔ service

RTT ≈1ms ●

distance ≈100km

Bandwidth

funnel of 1M’s of 1Gb/s connections

Resilience

make edge sites autonomous, minimize failure domain to 1 site

Regulations

keep sensitive data on-site / within regulatory region

Distribution

need to deploy and perform the lifecycle

management of distributed systems

Scale

need to manage 10-100k’s of sites with 1-10’s of nodes

each

Edge - 5G Constraints

48


RRC PDCP High-RLC

Low-RLC

High-MAC

Low-MAC

High-PHY

Low-PHY RF

RFRRC PDCP High-RLC

Low-RLC

High-MAC

Low-MAC

High-PHY

Low-PHY

EPC nGCEdge

UEs

“0” 1

R15 R15

DL

ULTotal RAN Latency (transmission + scheduling+ processing switching/queuing + fronthaul + midhaul + etc.)

Service specific targets:• eMBB: Enable 10-50ms RTT• URLLC: Enable 2-5ms RTT

MAC HARQ loop (3TTI-FH TWL)

2: PDCP-RLCLTE: 30 SCF.. 10ms5G FWA FH: 5-10ms

5: Split MACLTE: 6msSCF

5G FWA FH: 1-3ms

6: MAC-PHYLTE: 2msSCF

5G FWA FH: 100-500us

7-*: Split PHYLTE: .25-2ms5G FWA FH: 50us

8: BBU-RRH LTE: 250usSCF.. 100usCPRI

5G FWA FH: 50us

4: RLC-MACLTE: ?5G FWA FH: 1-3ms

3: Split RLCLTE: ? 5G FWA FH: 5-10ms

RAN targets [IMT2020]:• eMBB: ≤8ms RTT• URLLC: ≤2ms RTT• CP: ≤10ms3GPP

[CP time is for idle-act, multiple steps !]

Transport targets:• CPRI: 200us RTT• eCPRI: CoS “high” 100 us; CoS “med” 1ms• NGFI: TBD• 802.1CM: TBD (so far aligned with eCPRI)• SCF: 6ms (sub-ideal), 2ms (near ideal), 250us (Ideal)

Budgets split primarily to:• processing times, and• transport allocations• tradeoff, “flexible”…

Splits, Latency Budgets and Allocation Targets

49

Presenter

Presentation Notes

FWA = Fixed Wireless Access NGFI = Next Generation Fronthaul Interface OWD = One Way Delay SCF = Small Cell Forum??

https://www.itu.int/md/R15-SG05-C-0040/en

http://www.ieee802.org/1/pages/802.1cm.html


● DU, CU and vBBU – Containers attractive to deliver control functions and may be data plane functions in future

● 5G NG-Core - Separate Control and User Plane allows flexible deployment of functions○ => Orchestration models needed to place functions ○ => Granularity functions or services for flexibility – re-usability

● Support for Edge Computing and efficient UPF re-selection/redirection○ => Common orchestration of core and other functions○ => Reusability of components

● Support Network Slicing based on modular design and multi slice connectivity from UEs○ => Resource partitioning from air interface to edge to core

● Stateless functions – compute decoupled from storage● OTT Applications and offload

Red Hat is preparing the infrastructure so that these services can be delivered from it

vRAN and NG-Core - LTE+ and 5G

50


● Real Time requirements○ Latency sensitive => 1ms processing budget at DU node

● Timing synchronization○ vRAN applications require host and VM time synchronization with <1us accuracy

● Accelerator support○ Data plane accelerators for Split RAN architecture => FPGA and/or Smart NIC

● High throughput○ Dedicated CPU and thread pinning and other performance tuning

5G Requirements

51


● More cell sites and Edge compute => massive scale○ Distribution of compute => HA models○ Ex: 1.2M servers required for Edge compute and vRAN

● Massive Scale => Orchestration challenges○ Provisioning and life cycle management tools

● Hybrid Workloads○ Virtual Machines or Containers○ Any packaging model

● Storage Management○ At Edge and Core locations

5G Requirements

52


● Massive Distribution of sites – Scale

● Real time requirements & Hybrid Container/VM Models

● Separate Control and User Plane allows flexible deployment of functions

● Support for Edge computing with efficient UPF re-selection/redirection

● Support Network Slicing based on modular design and multi slice connectivity from UEs

● Stateless functions – compute decoupled from storage

5G Requirements, Summary

53

Support for Both Virtual Machines and Containers


Red Hat OpenShift Container Platform

● NFV started with VMs and is now evolving towards cloud native Network Functions based on containers

● Kubernetes is the leading Container Management Platform

● Red Hat OpenShift Container Platform (OCP) is built around Kubernetes

Seamless Transition To Containers

Presenter

Presentation Notes

Important to highlight that we've been contributing to Kubernetes since a long time….


with continued support of Virtual Machines

RED HAT OPENSTACK PLATFORM 13

OPENSHIFT CONTAINER PLATFORM 3.11

STANDARD HARDWARE

KVM Ironic (Bare Metal)

VMVM VM

App

App Networking

Networking

Kuryr*Octavia

App

App App App App App

Red Hat CloudForms

Red Hat Satellite

Red Hat OpenStack Platform Director

SHARED MANAGEMENT TOOLING

Red Hat Ansible Automation

Cinder

Containers Containers CinderIntegration

Compute Ceph

Seamless Transition To Containers

ObjectStorage

Registry

* Tech Preview


NETWORKING ANSIBLE NEUTRON ML2 PLUG-IN (GA in OSP 14)

57

Provisioning Network is configured in the switch

Boot BM on Tenant Network

ML2 Plug-in Configures

Switch

BM is Provisioned

ML2 Plug-in Configures

Switch

Tenant Network is configured in the switch

BM is ready

New Generic ML2 plug-in for bare metal multi-tenant based on Ansible Networking

L2 Switch

BMNIC

BMNIC

Multi Tenant Bare Metal as a Service - Ironic

https://blogs.rdoproject.org/2018/09/networking-ansible/

Presenter

Presentation Notes

The new networking ansible ml2 network shipped in OSP 14 covers the multi-tenant, non-trusted tenant environment where multiple tenants securely share the same pool of bare metal nodes. In this initial release we’ve tested it with Juniper/JunOS but Ansible and the driver abstract the functionality to any switch family supported by Ansible Networking. Community Announcement: https://blogs.rdoproject.org/2018/09/networking-ansible/ ML2 Ansible Networking epic ML2 Ansible Networking driver (BZ)

https://blogs.rdoproject.org/2018/09/networking-ansible/


OpenShif t Clust er

Master Node

Infra Node

Worker NodeDeploy OpenShift

RHOSP with Ironic

12

3

Deploy OpenStack with Ironic

OpenShift on Bare MetalDeploy OpenShift on RHOSP Ironic-managed bare metal nodes

RHOSP Directorintegrates openshif t -ansible

58


Example of OpenShift Usage ...

Presenter

Presentation Notes

Important to highlight that we've been contributing to Kubernetes since a long time….


http://images.engage.redhat.com/Web/RedHat/%7B17ba2edd-411d-4f72-a247-5c63416a6716%7D_OpenShift_OpenS tack_experiences_@_Proximus.pdf

Operators Are Experimenting ...

http://images.engage.redhat.com/Web/RedHat/%7B17ba2edd-411d-4f72-a247-5c63416a6716%7D_OpenShift_OpenStack_experiences_@_Proximus.pdf

Red Hat OpenStack Platform @ The Edge / NFV


● RHOSP 13 is already fit for some vRAN deployments○ Hyperconverged deployments + Virtualized Control Plane○ Real Time KVM○ Basic FPGA support with PCI passthrough○ Several PoCs and Trials performed with vBBU VNF partners on RHOSP

Red Hat OpenStack Platform @ The Edge


Problem statement:● System calls and Interrupts introduce spikes

that impact the average scheduling latency

Solution:● New Real Time Compute Role that introduce deterministic and low average scheduling

latency● Max scheduling latency is 33us, 24 hours latency testing with RHEL 7.5 (cyclictest) ● PTP up to the VM (very accurate clock synchronization): required by 5G and Terrestrial

satellite stations

Red Hat Openstack / NFV @ The Edge, Realtime KVM


● RHOSP 13 is already fit for some some vRAN deployments○ Hyperconverged deployments + V irtualized Control Plane○ Real Time KVM○ Basic FPGA support with PCI passthrough○ Several PoCs and Trials performed with vBBU VNF partners on RHOSP

● RHOSP roadmap is addressing additional vRAN requirements such as○ Distributed Compute Nodes○ Enhanced FPGA support (Cyborg)○ Precision Time Protocol (PTP) support

Red Hat OpenStack Platform @ The Edge


Stand-AloneLarge Clusters

Resource Node

Resource Node

Resource Node All-in-One

Small Clusters Distributed Nodes

Control Plane

Control Plane

Control Plane

Site A

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

Resource Node

...

Site E

Site F

Site G

Control Plane

Control Plane

Resource Node

Control Plane

Resource Node

typically 1 rack

Site B

Site I

OSP Director

RHOSP 15 Goal - Tech Preview

RHOSP @ The Edge - Deployment

65


OSP 16 Goal (15 Tech Prev)

Characteristics:

● Distributed HCI nodes● Min of 3 HCI nodes per site● Multiple Ceph Clusters● Cells & Location awareness● Deployment of BM at scale

OSP Director

Site B

Control Plane

Control Plane

Resource Node

Control Plane

Resource Node

typically 1 rack

Site A

Site C Site D

spine & leaf

at scale

RHOSP @ The Edge - Distributed Compute Nodes

66


● Life cycle management for accelerators● GPUs, FPGAs, ASICs, NPUs, … ● Vendor-Neutral, Hypervisor-Neutral● https://wiki.openstack.org/wiki/Cyborg

OpenStack - Cyborg Service

Key Services:● Nova Placement:

○ Compute Node = Resource Provider○ vCPUs, memory = Resource Classes○ Properties = Traits

● Cyborg: Devices contain Nested Resource Providers○ Accelerators: Resource Classes (E.g. ACCELERATOR_GPU)○ Properties: Traits

● User ask: 4 vCPUs, 12 GB RAM, FPGA with specific properties

67

Presenter

Presentation Notes

OpenStack Summit presentation: https://docs.google.com/presentation/d/1lANAeZ3n1eafyhaz9gYhAnQ2Mopd-I3E_OK5NNRXFq8/edit#slide=id.p24

https://wiki.openstack.org/wiki/Cyborg


https://zshisite.wordpress.com/2017/10/25/sync-your-cloud-with-ptp/

● Host: PTP supported in RHEL (linuxptp)○ Boundary and ordinary clocks○ PTP Hardware Clock on NIC

● Guest:○ PTP in Guest OS with hardware support

via PCI passthrough/SR-IOV

or○ vPTP support in KVM (kvm_ptp)

Precision Time Protocol - PTP

https://zshisite.wordpress.com/2017/10/25/sync-your-cloud-with-ptp/

Red Hat OpenShift Platform @ The Edge / NFV


● Fast Path for High-Throughput I/O○ SR-IOV○ DPDK support○ QoS○ Overlay/Non-Overlay

● Flexible Network Management○ Multiple Networks / Interfaces○ BYO IP address management○ Discontiguous IP address blocks

● Multicast● IPAM● Encrypted tunnels● Egress firewall● IPv6 / DHCPv6● Isolation / Multi-tenancy● ACL / Security Groups

70

● Kubernetes community is busy addressing basic NFV gaps○ Enhance platform awareness, multiple network interfaces,...○ Only recently created a new IoT/Edge working group

https://github.com/kubernetes/community/tree/master/wg-iot-edge

Critical Networking Feature For NFV Workloads

https://github.com/kubernetes/community/tree/master/wg-iot-edge


Feature(s): HugePages, CPU Manager, Device Manager

Description: We spoke about Device Manager here. CPU Manager Policy allows you to tell kube that your workload requires an affinity to a CPU core. Maybe your workload needs CPU cache affinity and can’t handle being bounced around to different CPU cores on the node via normal fair share scheduling on linux. HugePages allows you to request that your workload consume a specific amount of HugePages.

Performance Pods

How it Works: CPU manager is a flag on the kubelet that has the option of none or static. Static will cause guaranteed QoS pod access to exclusive CPU cores on the node. HugePages is a flag you set to true on the master and kubelet. The nodes will then be able to tell if any HugePages are available and workloads can request them via the pod definition.

ubelet device manager

CPU Manager Policy

# cat /etc/origin/node/node-config.yaml...kubeletArguments:...feature-gates:- CPUManager=truecpu-manager-policy:- staticcpu-manager-reconcile-period:- 5skube-reserved:- cpu=500m

Result:

# oc exec pod-name -- cat /sys/fs/cgroup/cpuset/cpuset.cpus2# oc exec pod-name -- grep ^Cpus_allowed_list /proc/1/statusCpus_allowed_list: 2

HugePages

# cat /etc/origin/node/node-config.yaml...kubeletArguments:...feature-gates:- HugePages=true

Pod spec:

resources:requests:cpu: 1memory: 256Mi

limits:cpu: 1memory: 256Mi

# cat /etc/origin/master/master-config.yaml...kubernetesMasterConfig:apiServerArguments:...feature-gates:- HugePages=true

Pod spec:

resources:limits:

hugepages-2Mi: 100Mi

Both the variable name and value are configurable.

OCP3.10

https://www.redhat.com/en/blog/paving-way-intelligent-and-performance-sensitive-applications-kubernetes-red-hat-openshift-container-platform-310

https://trello.com/c/REzRX4UW/876-node-promote-hugepages-to-supported

https://trello.com/c/5Pj2QkcP/877-promote-cpu-manager-pinning-to-supported

https://kubernetes.io/docs/concepts/cluster-administration/device-plugins/

https://docs.google.com/presentation/d/1p0DFiZAPyJiQNtFGGGIcfspHb9kLMJcElw1ShxlJFx0/edit#slide=id.g31380bb1cf_0_151

https://www.redhat.com/en/blog/paving-way-intelligent-and-performance-sensitive-applications-kubernetes-red-hat-openshift-container-platform-310


Container Network Interface (CNI)

● Problem: Kubernetes only supports one network interface, “eth0”, but we need:○ Functional separation of control/data planes○ Link aggregation for network redundancy○ Different network protocol stacks, capabilities, SLAs○ Traffic isolation / Network segregation and security○ QoS

● Solution: Multus “meta plug-in” for Kubernetes CNI● Enables multiple network interfaces per pod, each assigned a

different CNI plug-in defined in pod spec○ Each with its configuration defined in CRD objects

● SR-IOV enablement

Podeth0 eth1

Pod with Multus

openshif t -sdn CNI plug-in

SR-IOVCNI plug-in

Multus

CNI

...

Kubernetes

OCP4.0

72


● Method of packaging, deploying and managing applicat ions

● Encode human operat ional knowledge● Automat ically patch, upgrade,

recover, and tune apps and services● K ubernetes-native● Purpose-built for a specific application

or service

S DKLIFE CYCLE

MANAGE ME NTME TE R ING

ht tps:/ / github.com/ operator-framework

Kubernetes Operators

73

Presenter

Presentation Notes

IN RED HAT OPENSHIFT, we deliver automated operations with Kubernetes Operators. Operators are a method of packaging, deploying, and managing a specific Kubernetes applications. Kubernetes Operators encode the human knowledge required to install, upgrade / patch, recover from failure, and tune applications and services. Or more specifically: operators extend the Kubernetes API to create, configure, and manage instances of applications on behalf of a Kubernetes user. Operators execute these actions automatically, based on the state of your system - they check current state against desired state, and revert back to your desired state. They’re Kubernetes-native, and will run on any Kubernetes cluster. They’re designed to automate routine tasks away from your talent, and free them up to pursue more meaningful work. And operators are gaining traction: there’s over 50 community operators already, with several contributed by leading ISVs themselves (see: https://github.com/operator-framework/awesome-operators) === For more info on how Operators are built and managed, see the Operator Framework === FAQ: Are operators open source? Not always; the Operator Framework to *build* them is open source, but the resulting Operator may not be, depending on the entity that builds it Where can Operators run? Operators can run on any Kubernetes cluster, including OpenShift Where did the idea of Operators come from? CoreOS developed operators in 2016; as part of Red Hat, we’re able to incorporate that technology into OpenShift. Why should I pay Red Hat for the privilege of running Operators? Operators themselves have a lifecycle, and must be managed (though this can be done with the open source Operator Lifecycle Manager). You also need a way to centrally administer them. OpenShift provides a platform for central lifecycle and administration, along with The Operator Framework consists of three parts - but is all designed to provide a streamlined, uniform way of writing and managing Operators (which again, can be deployed to ANY Kubernetes cluster, not just OpenShift). Operator SDK (Software Development Kit) makes it easy to build or modify Operators for your own organizations - or, if you’re an ISV, for your particular offering - without specialized knowledge of the Kubernetes API. Focus more on building what your organization and customers need, not on the intricacies of Kubernetes. Operator Lifecycle Manager (OLM) manages the lifecycle of all Operators and their dependencies on your Kubernetes cluster. With OLM, you can install applications with a single kubectl command, and you’re not saddled with manually managing permissions or updates. Operator Metering This portion of the framework gives admins insight into usage for specific Operators. Usage metrics are unique to each Operator - metering watches custom resources and collects data, and performs calculations and reports on that data back to the administrator.


APPLICATIONS AND SERVICESISV OperatorsCustom Operators (built w/ Operator SDK)

PLATFORM AND CLUSTER MANAGEMENTAutomated updates for Kubernetes, monitoring, security, registry and more

LINUX HOST Atomic, over-the-air updates for Red Hat CoreOS

ANY INFRASTRUCTURE

* coming soon

Full-Stack Automated Operations in OpenShift*

74

Presenter

Presentation Notes

*Full-stack automated operations are coming in OpenShift 4.0. With the coming release of OpenShift 4.0, every layer of your Kubernetes application stack will have automated installation, updates, and recover capabilities, powered by Kubernetes Operators. OpenShift will be self-driving on any infrastructure, on-prem or in the cloud, virtualized or bare metal. OpenShift 4.0 is targeted for a March 2019 GA Linux OS: Users on Red Hat OpenShift can have in-place, automatic upgrades with the latest security patches when choosing Red Hat CoreOS (formerly CoreOS Container Linux) as their Linux operating system with OpenShift Platform and cluster management: Portions of the core platform will be managed by Kubernetes Operators, along with monitoring, security, networking, storage, and registry solutions in OpenShift Applications and services: 60+ ISVs have committed to build operators for their solutions, which will be available to run with cloud-like ease, but everywhere you choose to build and run. Customers also have the option of building their own custom operators for proprietary solutions, or to modify existing operators using the Operator SDK The result: A more secure, always up-to-date Kubernetes application platform, without the headaches of manual and serial upgrades or downtime.


60+ Cert if ied ISV Operators in Red Hat Early Access Program

ISV Operators

75

Presenter

Presentation Notes

OVER 60 ISVs have committed to build operators through the Red Hat Early Adopter Program. As we get closer to OCP 4.0, we’ll be able to share more information on expanded ISV Operator plans, along with how we’re surfacing them within OpenShift. But the point remains: while Operators can introduce cloud-like simplicity to running Kubernetes applications, they still need central administration that works with all of your technology investments. If you’re on OpenShift, we’ve done that legwork for you through our platform.


Kubernetes Cluster APIs

76

● Brings declarative, Kubernetes-style APIs for cluster creation, configuration & management.

● Implemented using Kubernetes Operators to scale, upgrade, heal, etc. clusters in fully automated, provider-agnostic manner.

● Integrates with provider-specific Machine Controllers and Machine Config Daemons for cloud, virt. & bare metal environments.

● Manage your clusters consistently independently of where/how it’s deployed.

● New working group under Kubernetes Cluster Lifecycle S IG

https://kubernetes-sigs.github.io/cluster-api/

https://kubernetes-sigs.github.io/cluster-api/


https://www.openstack.org/assets/presentation-media/5G-Cloud-Native-v4.pdfhttps://www.youtube.com/watch?v=pCkzJ bkQYLE

Cloud Native 5G Core - Affirmed

This is a public example of the work Red Hat does with its partners to push Cloud Nativeness into telco networks

https://www.openstack.org/assets/presentation-media/5G-Cloud-Native-v4.pdf

https://www.youtube.com/watch?v=pCkzJbkQYLE


CONTAINER INFRASTRUCTURE AND ORCHESTRATIONContainer Applicat ion and Kubernetes orchestrat ion as provided by OpenShift are becoming the standard for new applicat ions.

VIRTUALIZED WORKLOADSVirtualized Workloads are not going anywhere fast! Business reasons (cost, t ime to market) and technical reasons (older/ different operat ing system)

CONVERGING INFRASTRUCTUREHow can we bring these two worlds closer together?

What About Existing Workloads?

78

Presenter

Presentation Notes

Application containers coupled with Kubernetes orchestration are increasingly the standard for new applications. Existing virtual machines aren’t going anywhere: Business reasons (cost) Technical reasons Requirements for non-Linux operating system or custom kernel Current systems force us to treat containers and virtual machines separately. How can we bring these worlds closer together?


Technology enabling developer use of OpenShift Container Platform as a unified platform for building, modifying, and deploying applications residing in both containers and virtual

machines in a common, shared environment.

Add virtual machines to your OpenShift projects directly from the service catalog!

What Is Container Native Virtualization?

79

Presenter

Presentation Notes

Use case: “As a developer how do I bring my existing virtualized workloads into new applications I am building in OpenShift Container Platform?” Unified platform to build and deploy applications regardless of whether they reside in application containers or virtual machines� Represents continuation of Red Hat’s commitment to Kubernetes as the future of application orchestration and the common fabric across the Open Hybrid Cloud Supporting development of new microservice applications in containers that interact with existing virtualized applications. Combining existing virtualized workloads with new container workloads on the same platform, thereby making it easier to decompose monolithic virtualized workloads into containers over time.�


Container

Virtual Machine

ENTERPRiSE CONTAINER HOST

Physical Machine

Container

Resultant virtual machines are able to run side by side directly on the same OpenShift nodes as application containers.

{Pods

Together At Last*

80

* Dev Preview

Presenter

Presentation Notes

Teams, that require the ability to provision the resources that they need - whether VM or container-based - and tie them into their CI/CD pipelines.


● Integrates direct ly into exist ing Kubernetes clusters

● Uses a k8s-nat ive approach whenever possible

● Leverage Container Networking Interface (CNI),

Container Storage Interface (CSI), and other k8s-

nat ive integrat ions

Operat ing System

Kubernetes

Physical Machine

VM Pod Regular Pod

http://kubevirt.io

KubeVirt: The CNV Upstream

81

Presenter

Presentation Notes

The KubeVirt project is at the heart of CNV, but is not the only component. Drops onto existing clusters without complex host level modification. Does this by being implemented as a CustomResourceDefinition, CRDs allow expanding Kubernetes with additional top level API objects. Defines objects including VirtualMachines (persistent VM definition), VirtualMachineInstances (instantiation of a VM that exists while the VirtualMachine is Running), VirtualMachinePresets (like Flavors or Instance types in some ways) Extends Kubernetes so VMs can be scheduled alongside containers in a way that we can leverage native integrations.

http://kubevirt.io


VF12

flannel, calico, (optional) istio, ..

25 Gbps interface

Web App pod

serviceeth0

VF11

vSwitch

Network Function pod

serviceVF10 virtio eth0

VF13

vSwitch

KubeVirt pod

Port FWDvirtioeth0 VF14

Virtual Machine

k8s VF14virtioservice

Network Plumbing for All Use Cases (WIP)...

Virtual Machine

pod, hosting one or more containers

Red Hat And Industry Work


Open Source Communit ies

Red Hat is Plat inum sponsor and key contributor to many upstream projects used in or around ONAP (Ansible, Drools, OpenDaylight, Kubernetes, OpenStack, Linux); Edge architecture contributor.

Red Hat is plat inum sponsor, Board and TSC member; act ive in mult iple projects, with driving role in Edge use case & requirements analysis and VCO 1.0/ 2.0 PoCs.

Red Hat is leading OpenStack contributor, act ively part icipat ing in Edge WG to identify gaps, address in relevant OpenStack projects.

Red Hat is contributor to many IoT-related projects, with focus on Linux, cloud platform / infrastructure, and messaging technologies.Red Hat is a leading Kubernetes

contributor, init iated and leads IoT/ Edge WG to identify gaps, address in relevant Kubernetes SIGs.

Red Hat is Board and TSC member; tech lead for the Kubernetes-Native Infrastructure Blueprint Family

IoT Use Cases

Telco Use Cases (e.g. VCO, vRAN)

Edge Community Involvement


Standards Development Organizat ions (SDO’s) and Industry Alliances

Red Hat is:• NFV ISG member• MEC ISG member• ZSM ISG member

Red Hat is:• virtual RAN fronthaul project member• Open RAN project member• Edge Computing project member

Red Hat is:• act ively contribut ing to ARM-enablement• sponsor of and lead contributor to Fedora IoT,

Linaro EDGE’s reference platform

Red Hat is:• ORAN Member since opened to vendors (Dec 2018) • Focus on WG6 (cloudificat ion & Virtualizat ion) WG• Was previously member of CMCC’s CRAN consort ium which was

merged with xRAN.org to form O-RAN

Edge Community Involvement


https://www.o-ran.org/resources/

WP6: Cloudification and MANO enhancements (AT&T, Orange)

O-RAN Alliance Reference Architecture

86

Presenter

Presentation Notes

NFV Strategy: 100% virtualized, 100% software, i.e. get rid of SR-IOV 100% software virtualization stack - Challenge is the performance HW dependency creates huge testing matrix for VNFs: TEF, Verizon Agility/Velocity in new features development Strategy: provide a DPDK based vSwitch providing virtio interfaces Get the door open to HW offload as long as it’s transparent/virtio-standard from VM/container perspective Go live with OVS-DPDK and migrate to VPP Improve DPDK/virtualization performances and features (benefit to any DPDK based vSwitch) Note: VM or containers have the same vSwitch needs, and we provide them virtio interfaces

https://www.o-ran.org/resources/


Akraino Edge Stack is an open source software stack that improves the state of edge cloud infrastructure for carrier, provider, and IoT networks. https://www.akraino.org/

Akraino Edge Stack

https://www.akraino.org/


Red Hat is project lead of Akraino’s “Kubernetes-Native Infrastructure for Edge” (KNI-Edge) blueprint family. It’s blueprints

● implement the Kubernetes Cluster API for declarativecluster deployment, configuration, and lifecycle manage-ment in a provider-agnostic manner,

● leverage the Operator Framework for automated andsecure edge stack lifecycle management, and

● optimize for Kubernetes-native container workloads, butallow mixing in VM workloads via KubeVirt as needed.

Edge sites are deployed and upgraded in a rolling manner through a logically centralized CI/CD system (both software and configuration changes).

...

App1Op App2Op App2Op ...

baremetalprovider

OpenStackprovider

AWSprovider

...

Machine & Machine Config APIs / Operators

Cluster API / Operator

Akraino KNI-Edge Blueprint Family

88


IoTGATEWAY

EDGE

CONNECTED “THINGS”

Edge analytics

Machine learning

Telemetry

Telemetry

Management

IoT INTEGRATION

HUB

Fedora Internet of Things Edition

IoT GATEWAY EDGE IoT DEVICE EDGE PROJECTS

PRODUCTS

• HW enablement (e.g. ARM)• OSTree (updates)• Ignit ion (enrollment)• Ansible (management)• Kubernetes (containerizat ion)

• Distributed Computing and Edge SKUs

• Containerized applicat ions• Centralized managementIoT messaging

89

IoT Gateway And Device Edge

Red Hat Telco Customers & Ecosystem

Presenter

Presentation Notes

We don’t have a single strategy as Edge depends on 3 main parameters: we have one strategy per combination 1/ deployments 2/ form factor 3/ workloads: VMs/Containers/xxx


SOME OF OUR TELCO CUSTOMERS

● Logos above are only a subset of the public Telco references (OpenStack & OpenShift)● It can cover NFV but also Internal IT or Public Clouds; some through Partners (NEP/SI)● Telcos are in prod with OpenStack; experimenting OpenShift for NFV● Some Telcos are in prod with OpenShift for non-NFV offers

...

91

Presenter

Presentation Notes

VF = Vodacom (through Cisco) NTT East (through Cisco)


Management &Orchestration

(MANO)

Compute Storage Network• OSP Director• CloudForms• Satellite

vDDoS vRoutervCPE vCDNvEPC vMessagingvPCEF/RFvLB vADC vIMSvSBCClassifier

SMS

MM

SipM

essaging

Various NEP VNFs

Red Hat NFV Partner Ecosystem Representative

NFV Infrastructure (NFVI) + Virtualized Infrastructure Manager (VIM)

http://www.pcworld.com/article/2100120/mwc-dell-red-hat-team-up-on-network-virtualization.html


http://www.cisco.com/c/dam/en/us/td/docs/unified_computing/ucs/UCS_CVDs/ucs_rhos.pdf




http://pressoffice.telefonica.com/jsp/base.jsp?contenido=/jsp/notasdeprensa/notadetalle.jsp&id=0&origen=portada&idm=eng&pais=1&elem=20833

http://pressoffice.telefonica.com/jsp/base.jsp?contenido=/jsp/notasdeprensa/notadetalle.jsp&id=0&origen=portada&idm=eng&pais=1&elem=20833





#TTO #vRAN #Edge #5G93 Red Hat Confidential

NEP Ecosystem


● Supportability - Partner supported in production for VNF and datapath (SR-IOV, OVS, OVS-DPDK)● Install and Inservice update - verify end-to-end install with RHOSP; verify “yum update” works flawlessly; ● Security - use the right drivers for secure VM-to-NIC, VM-to-VM communication● Tools are certification tests run by the partner, and then reviewed by Red Hat

○ Step one: architectural review for the VNF by PM/Partner team ○ Step two: Workflow guide https://access.redhat.com/documentation/en-us/red_hat_openstack_certification/1.0/html/red_hat_openstack_application_and_vnf_workflow_guide/

● Red Hat Openstack Platform application and VNF Policy Guide:https://access.redhat.com/documentation/en-us/red_hat_openstack_certification/1.0/html/red_hat_openstack_application_and_vnf_policy_guide/

Partner needs to recertify with each long lifecycle RHOSP release and major release of RHEL● Provide early content and ask partner to validate at least 1 month before GA through High Touch Beta● VNF lab upcoming - Automated certification/validation process with every release, partner encouraged to participate● Find certified or validated VNFs : Certified/Validated VNFs● RHEL Network Adapter support https://access.redhat.com/articles/1390483● Hardware certification: https://access.redhat.com/ecosystem/search/#/ecosystem/Red%20Hat%20OpenStack%20Platform?category=Server

Additional information for planning purposes● Network Functions Virtualization Planning and Prerequisites Guide: https://access.redhat.com/documentation/en-

us/red_hat_openstack_platform/12/html/network_functions_virtualization_planning_and_configuration_guide/

● Networking Guide for RHOSP13: https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/networking_guide/

Value Of Certification / Validation / Integration

https://access.redhat.com/documentation/en-us/red_hat_openstack_certification/1.0/html/red_hat_openstack_application_and_vnf_workflow_guide/

https://access.redhat.com/documentation/en-us/red_hat_openstack_certification/1.0/html/red_hat_openstack_application_and_vnf_policy_guide/

https://access.redhat.com/ecosystem/search/#/ecosystem/Red%20Hat%20OpenStack%20Platform?sort=sortTitle%20asc&tags=VNF&category=Software

https://access.redhat.com/articles/1390483

https://access.redhat.com/ecosystem/search/#/ecosystem/Red%20Hat%20OpenStack%20Platform?category=Server

https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/12/html/network_functions_virtualization_planning_and_configuration_guide/

https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/networking_guide/

plus.google.com/+RedHat

linkedin.com/company/red-hat

youtube.com/user/RedHatVideos

facebook.com/redhatinc

twitter.com/RedHatNews

Documents

Towards vRAN / Edge / 5G vRAN... · data processing compute nova networking cinder glance swift neutron ironic storage block image object bare-metal provisioning horizon tripleo dashboard