Slide 1

Slide 1 text

PPJ #6 Virtualization & Software Defined Networking http://eueung.github.io/EL5244/ Dr.-Ing. Eueung Mulyana | 2015H2

Slide 2

Slide 2 text

This material is mainly a derivative and remix work. Most of the texts and illustrations are taken from the talks/lectures given by the referenced networking professors/gurus/ninjas (Credits at the end of the Slide).

Slide 3

Slide 3 text

SDN @ Telcos + ISPs

Slide 4

Slide 4 text

Now ISPs and Telcos are Starting to Respond McKeown

Slide 5

Slide 5 text

The Problem • Global IP traffic growing 40-50% per year • End-customer monthly bill remains unchanged • Therefore, cost of ownership needs to reduce 40-50% per Gb/s per year • But in practice, reduces by <20% per year $30 /month time Revenue

Slide 6

Slide 6 text

On SDN “Because there is no army that can hold back an economic principle whose time has come.” John Donovan, AT&T (2014) Cisco 2014 • #1 Competitor: A software company (VMware) • #1 Fear: Baremetal switches

Slide 7

Slide 7 text

The Perspective … Search Facebook YouTube Netflix Amazon Network: Baremetal xChip/x86 + Linux (500 thousand/yr) Compute: Baremetal x86 + Linux (10 million/yr) Public Internet Telcos/Cable Open Connect B4 Swan Google Fiber Google WiFi ARM + Linux Baremetal xChip/x86 + Linux/BSD

Slide 8

Slide 8 text

Baremetal + Linux + Apps Reduce capital cost Move to baremetal switches + Linux Reduce operational cost Streamline protocols, features and software Increase price Differentiate with new proprietary software services

Slide 9

Slide 9 text

9 Server Switch Radio VM Linux Linux Linux Linux Programmable Hardware Open Source Software PIC Linux Remote Control Plane Routing, Traffic Engineering, Mobility, VPN Proprietary Software DPI, Load Balance, SPAM New Services Overall System Management

Slide 10

Slide 10 text

(Core) Transport Networks

Slide 11

Slide 11 text

SDN – The Promise • Simplified Operations • Platform Consolidation • Savings in Application and HW • The “SDN” matra • The ability of SW to program the network’s forwarding plane in real-time • Rapid Service Definition and Deployment • Service Automation • Viable SW Sources • Separation of features from HW realization • Network Apps  NFV • Resource Sharing  Hypervisor Virtualization Service Agility Reduce Cost Programmability

Slide 12

Slide 12 text

Virtualization – NFV & Hypervisors Function Virtualization • Rework of network apps to urn on arbitrary HW, independent of custom networking HW Resource Virtualization • Functionality to logically abstract a single set of network resources and make them usable as independent networks Separation of “what” from “where”

Slide 13

Slide 13 text

Programmability Present Mode of Operation • Forwarding plane is under the control of embedded control plane and device provisioning SDN Mode of Operation • Forwarding plane is programmable in real- time by independently developed software

Slide 14

Slide 14 text

Programmable Optical Networks

Slide 15

Slide 15 text

Programmable Optical Networks • Direct Model envisions SDN API’s placed on every optical switching element • Exposes SDN controller to analog complexities of ON • Only viable in simple networks • Overlay Model abstracts the optical network and presents it as simple elements to the controller • Hides analog optical complexities from controller and apps Finding the appropriate level of abstraction is key to virtualization

Slide 16

Slide 16 text

Virtualization in Optical Networks

Slide 17

Slide 17 text

Virtualization in Optical Networks • Overlay Model can be extended to abstract the network into multiple networks, for multiple controllers • Allows tenant controllers to control virtualized slices of the carrier network • Function Virtualization is not a hot topic in ON Network Hypervisor creates programmable and virtualized ON

Slide 18

Slide 18 text

Where Transport Fits in SDN Model • Hypervisors allowed storage and compute resources to be managed together • SDN allowed networking to join the fun through an orchestration layer

Slide 19

Slide 19 text

Where Transport Fits in SDN Model • Apps can be written on top of orchestration layer • Transport is added by extending the SDN controller function • A Network hypervisor virtualized the transport network and presents an abstracted representation to the SDN controller • All network layers are included in the control model  end-to-end new apps are possible

Slide 20

Slide 20 text

Where Transport Fits in SDN Model The Network Hypervisor has a big job in managing all transport

Slide 21

Slide 21 text

Application Drivers in ONs • Use Cases • DC Connectivity • Cloud Bursting • Secure Multi-Tenancy • Optical VPN • Multi-Layer Optimization SDN turns the network into a programmable resource

Slide 22

Slide 22 text

SDN in the Core … • SDN turns the network into a programmable resource • Network Hypervisor makes optical layer programmable • Resource Virtualization: Network Hypervisor abstracts the network into slices

Slide 23

Slide 23 text

Credit • Scott Shenker, The Future of Networking and the Past of Protocols • Nick McKeown, Stanford University, Many Talks/Articles • Jennifer Rexford, COS 597E, Princeton University • ADVA Optical Networking, Many Talks/Slides/Articles