IT1634 – SDN Unit 1.pptx Software defined

C. Raj Kannan,AP/IT
IT1634 – Software Defined Networks
UNIT I - INTRODUCTION TO SDN

UNIT I - INTRODUCTION TO SDN
History of Software Defined Networking (SDN) – Modern
Data Center – Traditional Switch Architecture – Why SDN –
Evolution of SDN – How SDN Works – Centralized and
Distributed Control and Date Planes

Software Defined Networks
 Over the past two decades, networks have come under increased
traffic demands and increased scrutiny as both organizations and
consumers increasingly rely on network connectivity for sales,
customer service, internal communications and document sharing.
 Traditional network architectures are not designed in a way that
meets current requirements. SDNs offer an alternative paradigm for
meeting the needs of users, companies and service providers.
 Software-Defined Networking (SDN) is an approach to
networking that uses software-based controllers or application
programming interfaces (APIs) to communicate with underlying
hardware infrastructure and direct traffic on a network.

 Software-Defined Networking (SDN) is a network architecture
approach that enables the network to be intelligently and centrally
controlled, or programmed,’ using software applications.This helps
operators manage the entire network consistently and holistically,
regardless of the underlying network technology.
 Software-defined networking is an architecture designed to make a
network more flexible and easier to manage.
 SDN centralizes management by abstracting the control plane from
the data forwarding function in the discrete networking devices.

 SDN is important because it gives network operators new ways to
design, build and operate their networks.
 Software-defined networking paired with network functions
virtualization is a key technology needed to meet new demands.
 SDN separates the network’s control and forwarding planes and
provides a centralized view of the distributed network for more
efficient orchestration and automation of network services.
 The SDN controller platforms that organizations use allow for
communication between the now separated network planes.

Historical Background
• The major communications networks around the world in the
first half of the 20th century were the telephone networks
• Composed of switching offices, each of which was connected to
thousands of telephones
• Switching offices were, in turn, connected to higher-level switching
offices (toll offices), to form a national hierarchy
• The vulnerability of the system was that the destruction of a few
key toll offices could fragment it into many isolated islands
6

Historical Background
• Paul Baran, a Polish immigrant who became a
researcher working at Rand Corporation in the US
around 1960, argued that in the event of enemy
attack networks like the telephone network were
easy to disrupt
• Mr. Baran’s proposed solution was to transmit the
voice signals of the phone conversations in packets
of data that could travel autonomously – survivable
networks (1964)1
• Digital packet-switching technology
7
1. P. Baran, Baran, Paul, “On Distributed Communications: I. Introduction to Distributed Communications
Networks,” RAND Corporation, 1964. https://www.rand.org/pubs/research_memoranda/RM3420.html

Legacy Networks Overview
• A network calledARPANET eventually was implemented using
Baran’s ideas
• Funded by the U.S.Advanced Research Projects Agency (ARPA)
• This decentralized, connectionless network grew over the years
until bursting upon the commercial landscape around 1990 in the
form of the Internet
• The Internet was a distributed, connectionless architecture
8 1972
1969
1977

Legacy Networks Overview
• In the early days, existing protocols were not suitable for running
over different networks
• In 1974,TCP/IP model and protocols were invented by Robert
Khan andVinton Cerf1
9
1. V. Cerf, R. Kahn, “A Protocol for Packet Network Intercommunication,” IEEE Trans. on Comms, vol. 22,

CSNET and NSFNET
• In 1981, the National Science Foundation (NSF) established the
Computer Science Network (CSNET) to provide connect (to
ARPANET and other networks) to all university computer
scientists
• In 1985, NSF established the NSFnet to link together five
supercomputer centers that were then deployed across the U.S.
10
NSFNET backbone
Backbone
Campus
networks
Regional
networks
Regional
networks
Campus
networks
Campus
networks

Differences between Traditional Networking and SDN

The Modern Data Center
• In 1991, NSFNET lifted its restrictions on the use of NSFNET
for commercial purposes
• NSFNET itself would be decommissioned in 1995, with Internet
backbone traffic being carried by commercial Internet Service
Providers (ISPs)
• The main event of the 1990s was to be the emergence of the
WorldWideWeb
• Invented at CERN byTim Berners-Lee between 1989 and 1991
• The web brought the Internet into the homes, businesses,
millions of people
12

• A number of companies emerged as big winners in the Internet space
• Microsoft, Cisco,Yahoo, e-Bay, Google,Amazon
• The web gave rise to data centers, hosting heavily subscribed web services
• Servers were physically arranged into highly organized rows of racks of servers
• Racks were hierarchically organized such thatTop-of-Rack (ToR) switches provided
the networking within the rack and the inter-rack interface capability
13

• A modern physical servers can host hundreds of virtual machines
(VMs), results in thousands (or even millions) of VMs communicating
within the datacenter
• These VMs are now communicating via a set of protocols and devices
that were optimized to work over a large, disparate geographical area
with unreliable links
• While still important, survivability was not that relevant (in contrast
to 1970s, 1980sWANs) in the emerging data center
• Network management systems designed for carrier public networks or
large corporate intranets simply cannot scale to these numbers
• A new network management paradigm was needed
14
While the modern data center was the premier driver behind the SDN fervor,by no
means is SDN only applicable to the data center

Traditional Switch
Architecture
15

• The data plane consists of the various ports that are used for the reception and
transmission of packets and a forwarding table with its associated logic
• The data plane assumes responsibility for packet buffering, packet scheduling, header
modification, and forwarding
• If an arriving packet’s header information is found in the forwarding table, it may be
forwarded without any intervention of the other two planes
16
Data, Control, and Management Planes

• Not all packets can be handled exclusively at the data plane, sometimes simply
because their information is not yet entered into the table, or because they belong to
a control protocol that must be processed by the control plane
• The main role of the control plane is to keep current the information in the
forwarding table so that the data plane can independently handle as many packets as
possible
17

• Network administrators configure and monitor the switch through the management
plane
• The management plane extracts information from or modifies data in the control
and data planes as appropriate
• The network administrators use some form of network management system to
communicate with the management plane in a switch (e.g., command-line interface)
18

• When a packet arrives on an interface, it is forwarded to the
control plane where the CPU matches the destination address
with an entry in its routing table
• The router does this for every packet
19
Software-based Routing and Bridging
Control Plane
Data Plane
CPU
Ingress
interfac
e
Egress
interfac
e

• The first major use of hardware acceleration in packet switching was via the
use ofApplication-Specific Integrated Circuits (ASICs) for table look-ups
• In the mid-1990s advances in Content-Addressable Memory (CAM)
technology made it possible to perform very high speed look-up using
destination address fields
20
Hardware Look-up of Forwarding Tables
Control Plane
Data Plane
CPU
Ingress
interfac
e
Egress
interfac
e
ASIC

EVOLUTION OF SWITCHES AND CONTROL PLANES

EVOLUTION OF SWITCHES AND CONTROL PLANES
 Simple Forwarding and Routing Using Software
 Independence andAutonomy in Early Devices
 Software Moves Into Silicon
 Hardware Forwarding and Control in Software
 The Growing Need for Simplification
 Moving Control Off of the Device

Simple Forwarding and Routing Using Software
 In the early days of computer networking, where almost everything
other than the physical layer (layer one) was implemented in
software.
 Whether the devices were bridges, switches, or routers, software
was used extensively inside the devices in order to perform even the
simplest of tasks, such as MAC-level forwarding decisions.
 This remained true even through the early days of the
commercialized Internet in the early 1990s.

Independence and Autonomy in Early Devices
 Early network device developers and standards-creators wanted each
device to perform in an autonomous and independent manner, to the
greatest extent possible.
 This was because networks were generally small and xed, with large
ﬁ
shared domains.
 Developers went to great lengths to implement this distributed
environment with intelligence resident in every device.
 Whenever coordination between devices was required, collective
decisions could be made through the collaborative exchange of
information between devices.
 Interestingly, many of the goals of this distributed model, such as
simplicity, ease-of-use, and automatic recovery, are similar to the goals of
SDN.

Software Moves Into Silicon
 Today, switching devices are typically composed of hardware components
such as Application Speciﬁc Integrated Circuits (ASICs),Field-Programmable Gate
Arrays (FPGAs),andTernary Content Addressable Memories (TCAMs).The combined
power of these integrated circuits allows for the forwarding decisions to be
made entirely in the hardware at line rate.
 This has become more critical as network speeds have increased from one
Gbps to ten Gbps, to forty Gbps, and beyond.
 The hardware is now capable of handling all forwarding, routing, Access
Control List (ACL),and QoS decisions.Higher-level control functions,
responsible for network-wide collaboration with other devices, are
implemented in software.
 This control software runs independently in each network device.

Hardware Forwarding and Control in Software
 Bridging (LayerTwo Forwarding)
 Basic layer two MAC forwarding of packets is handled in the hardware tables.
 Routing (LayerThree Forwarding)
 In order to keep up with today’s high-speed links and to route packets at link
speeds, layer three forwarding functionality is also implemented in hardware
tables.
 Advanced Filtering and Prioritization
 General traf c management rules, such asACLs, which lter, forward, and
ﬁ ﬁ
prioritize packets, are handled via hardware tables located in the hardware (e.g.,
inTCAMs), and accessed through low-level software.
 Control
 The control software used to make broader routing decisions and to interact
with other devices in order to converge on topologies and routing paths is
implemented in software that runs autonomously inside the devices.

The Growing Need for Simplification
 Attempting to provide simplicity by adding features to legacy
devices tends to complicate implementations rather than
simplifying them.
 In addition to simplifying the devices themselves, there is an
opportunity to simplify the management of the networks of
these devices. Rather than using primitive network
management tools such as SNMP and CLI, network operators
would prefer to use policy-based management systems. SDN
may enable such solutions

Moving Control Off of the Device
 SDN attempts to segregate network activities in the following manner:
 Forwarding, Filtering, and Prioritization
 Forwarding responsibilities, implemented in hardware tables, remain on the
device.
 Control
 Complicated control software is removed from the device and placed into a
centralized controller, which has a complete view of the network and the ability
to make optimal forwarding and routing decisions.
 Application
 Above the controller is where the network applications run, implementing
higher-level functions and, additionally, participating in decisions about how
best to manage and control packet forwarding and distribution within the
network.

Cost
 Increased Cost of Development
 Despite the overall downward trend in the cost of networking
hardware, this growing complexity acts as an upward pressure on the
hardware component costs due to the processing power required to run
that advanced software as well as the storage capacity to hold it.
 Closed Environments EncourageVendor Lock-in
 With many vendors adding such enhancements, the end result is that
each vendor product will have dif culty interoperating smoothly with
ﬁ
products from another vendor.
 Complexity and Resistance to Change
 Increased Cost of Operating the Network

Cost
 Complexity and Resistance to Change
 The ideal would be a simpler, more progressive world of networking,
with open, efﬁcient, and less expensive networking devices.
 Increased Cost of Operating the Network
 As networks become ever-larger and more complex, the Operational
Expense (OPEX) of the network grows.
 This component of the overall costs is increasingly seen to be more
signiﬁcant than the corresponding Capital Expense (CAPEX) component.
 SDN has the capacity to acceleratethe automation of network management
tasks in a multivendor environment

SDN IMPLICATIONS FOR RESEARCH AND INNOVATION
 Status Quo Benefits IncumbentVendors
 The small players will struggle to survive, attempting to chip away at
the industry giants, but with limited success, especially since the proﬁt
margins of those giants are so large.
 SDN Promotes Research and Innovation
 A number of universities collaborated to propose a new standard for
networking called OpenFlow, which would allow for this free and open
research to take place.
 This makes one wonder if SDN will ultimately be to the world of
networking what Linux has become to the world of computing.

DATA CENTER INNOVATION
 Compute and StorageVirtualization
 These technological advancements allow servers and storage to be
manipulated quickly and efficiently.While these advances in computer
and storage virtualization have been taking place, the same has not been
true in the networking domain
 Inadequacies in NetworksToday
 SDN holds the promise that the time required for such network
reconfiguration be reduced to the order of minutes, such as is already
the case for reconfiguration ofVMs.

DATA CENTER NEEDS
 Automation - Automation allows networks to come and go at will,
following the movements of servers and storage as needs change.
 Scalability -With data centers and cloud environments, the sheer
number of end stations that connect to a single network has grown
exponentially.
 Multipathing - the network must make optimal use of its resources,
and it must be resistant to failures of any kind
 Multitenancy - the idea of hosting dozens, or even hundreds or
thousands of customers or tenants in the same physical data center has
become a requirement
 NetworkVirtualization -The general idea of virtualization is that you
create a higher-level abstraction that runs on top of the actual physical
entity you are abstracting.

THE EVOLUTION OF NETWORKING TECHNOLOGY
 Mainframe Networking: Remote Terminals - Even in the age of
mainframes, remote connectivity to the mainframe was needed.
 Peer-to-Peer Point-to-Point Connections - In the point-to-point
connections, the network was trivial, with only the two parties
communicating with each other
 Local Area Networks - a way to connect the devices in order to allow
them to share information and collaborate
 Bridged Networks - these bridges were implemented in such a way
that each device was able to operate independently and autonomously
without requiring any centralized intelligence.
 Routed Networks - This was another application of autonomous
devices utilizing distributed protocols in order to allow each to make
appropriate forwarding decisions.

LEGACY MECHANISMS EVOLVE TOWARD SDN
 The capabilities of legacy switches were sometimes extended to
support detailed policy con guration related to security, QoS and
ﬁ
other areas.
 OldAPIs were extended to allow centralized programming of these
features.
 Some SDN providers have based their entire SDN solution on a rich
family of extendedAPIs on legacy switches, orchestrated by a
centralized controller.

SOFTWARE DEFINED NETWORKING IS BORN
 THE BIRTH OF OPENFLOW
 OpenFlow is a protocol speciﬁcation that describes the
communication between OpenFlow switches and an OpenFlow
controller.
 In reality, the term SDN did not come into use until a year after
OpenFlow made its appearance on the scene in 2008, but the existence
and adoption of OpenFlow by research communities and networking
vendors marked a sea change in networking, one that we are still
witnessing even now.
 Indeed, while the term SDN was in use in the research community as
early as 2009, SDN did not begin to make a big impact in the broader
networking industry until 2011.

SOFTWARE DEFINED NETWORKING IS BORN
 OPEN NETWORKING FOUNDATION
 By 2011 OpenFlow had gatheredenough momentum that the
responsibility for the standard itself moved to the Open Networking
Foundation (ONF).
 The ONF was established in 2011 by DeutscheTelekom,Facebook,
Google,Microsoft,Verizon, andYahoo!.
 It is now the guardian of the OpenFlow standard, and consists of a
number of councils, areas and working groups.
 One novel aspect of the ONF is that corporate members of the Board
of Directors consist of major network operators,and not the networking
vendors themselves.

SUSTAINING SDN INTEROPERABILITY
 Plugfests: Plugfests, staged normally at conferences, summits, and
congresses, are environments where vendors can bring their devices and
software in order to test them with devices and software from other vendors.
 Interoperability Labs: Certain institutions have built dedicated test labs
for the purpose of testing the interoperability of equipment from various
vendors and organizations. experimental devices and controllers from open
source contributors.
 Certi cation Programs:
fi There is a need for certi cation of switches so
fi
buyers can know they are getting a switch that is certified to support a
particular version(s) of OpenFlow.
 Education and Consulting: A complex, game-changing technological
shift such as that represented by SDN will not easily permeate a large
industry without the existence of an infrastructure to train and advise
networking staff about the migration.

OPEN SOURCE CONTRIBUTIONS
 The Power of the Collective
 In the world of software, it is possible for small players to develop
technology and make it freely available to the general public.
 The Danger of the Collective
 Open source software must undergo tests and scrutiny by even larger
numbers of individuals than its commercial counterpart.
 Open Source Contributions to SDN
 Huge advances in SDN technology are attributable to open source
projects.
 Multiple open source implementations of SDN switches, controllers
and applications are available.

NETWORK VIRTUALIZATION
 Network virtualization, in essence, provides a network service that
is decoupled from the physical hardware below that offers a feature
set identical to the behavior of its physical counterpart.
 An important and early approach to such network virtualization was
the Virtual Local Area Network (VLAN).
 VLANs permitted multiple virtual local area networks to co-reside
on the same layer two physical network in total isolation from one
another.

Plane Separation
• The first fundamental characteristic of SDN is the separation
of planes
• Data plane, implemented in the device
• Control plane, implemented by a centralized controller
45
Traditional networks SDN networks
W. Stallings, “Foundations of Modern Networking: SDN, NFV, QoE, IoT, and Cloud” Addison Wesley, 2017.

Plane Separation – Data Plane
• The data plane implements forwarding functionality (logic
and tables for choosing how to deal with incoming packets)
• Forwarding based on MAC address, IP address,VLAN ID, etc.
• The data plane may forward, drop, consume, transform,
replicate an incoming packet
46

Plane Separation – Data Plane
• It determines the correct output port by performing a lookup
in the address table in the ASIC (very high-speed hardware,
operating at terabits per second)
• Special-case packets (e.g., routing advertisements) that require
processing by the control plane are passed to that plane
47

Plane Separation – Control Plane
• The algorithms used to program the data plane reside in the
control plane
• Many protocols / algorithms require global knowledge (for
example, OSPF, BGP)
• The control plane is moved off of the switching device, onto a
centralized controller
48

SDN Operation
• Basic components (bottom-up)
• SDN switches (e.g. Open vswitches)
• Controller (e.g., ONOS controller)
• Applications (e.g., OpenFlow, forwarding)
49
Global view
app
BGP app
IDS app Business app
SDN
Controller
Northbound (REST,
JSON)
Southbound
(OpenFlow)
Control plane
Data plane
Flow table

SDN Operation – Switches
• SDN devices contain forwarding functionality
• Forwarding information is stored in flow
tables
• The flow tables reside on the network device
and consist of a series of flow entries and
actions to perform when a packet matches an
entry
• If the SDN device finds a match, it takes the
appropriate configured action (e.g. forward)
• If it does not find a match, it can either drop
the packet or pass it to the controller
50
Global view
app
BGP app
SDN
Controller
Northbound (REST,
JSON)
Southbound
(OpenFlow)
Control plane
Data plane
Flow table

SDN Operation – Controller
• SDN controller implements control plane
functionality
• It presents an abstraction of the network to
the SDN applications running above
• It allows the SDN application to define flows
on devices and to help the application to
respond to packets which are forwarded to
the controller by devices
• It maintains a view of the entire network
(global network view)
51
Global view
app
BGP app
SDN
Controller
Northbound (REST,
JSON)
Southbound
(OpenFlow)
Control plane
Data plane
Flow table

SDN Operation – Applications
• SDN applications are built on top of the
controller
• Software applications can implement
forwarding, routing, overlay, multipath,
access control, etc.
• The application is driven by events coming
from the controller and from external inputs
• External inputs could include network
monitoring systems, Netflow, IDS, or BGP
peers
52
Global view
app
BGP app
SDN
Controller
Northbound (REST,
JSON)
Southbound
(OpenFlow)
Control plane
Data plane
Flow table

Flow Tables
• Flow tables are the fundamental data structures in an SDN device
• They allow the device to evaluate incoming packets and take the
appropriate action
• Flow tables consist of entries, each of which has match fields and actions
• OpenFlow explicitly specifies protocol headers on which it operates /
matches
53

IT1634 – SDN Unit 1.pptx Software defined

More Related Content

Similar to IT1634 – SDN Unit 1.pptx Software defined (20)

More from ssuser000e54 (19)

Recently uploaded (20)

IT1634 – SDN Unit 1.pptx Software defined