1) Requirements from SDN Approach

• Adaptability: networks will rapidly adapt and react to requirements, depend upon corporate policy
and network conditions.
• Automation: Policy modifications need to be propagated continuously in order to so it may be
possible to reduce manual work and errors.
• Maintainability: New features and capabilities are introduced (software upgrades, patches) should
be smooth and operating disruption minimal.
• Model management: the software for Network Management should enable network management
at the model level, instead of re configuring the network elements to implement conceptual
changes.
• Mobility: Mobility must be controlled including mobile user devices and virtual servers.
• Integrated security: Network applications have to integrate seamless security as a basic service
rather than as an add-on.
• On-demand scaling: Implementations must be equipped to scale up or down the network and its
facilities in support of on-demand request.

2) Characteristics of Software-Defined Networking (SDN)

• Separation of Control and Data Planes:
Traditionally, both control and data functions reside in the same device. SDN separates them, so
forwarding devices handle only data, while decisions are made centrally. This simplifies network
hardware and enhances flexibility.
• Centralized Control:
A single SDN controller manages the entire network, making decisions like routing and access
control. This unified control helps optimize traffic flow, improves visibility, and simplifies network
management and troubleshooting.
• Programmability:
SDN allows network administrators to program and automate behavior via software applications.
This enables rapid deployment of new services, dynamic adjustment of traffic, and consistent policy
enforcement across the entire network.
• Flexibility and Agility:
Network behavior can be changed on the fly without manual reconfiguration of devices. This makes
it easier to adapt to new business needs, deploy updates, or respond to faults and traffic shifts
quickly.
• Global View of the Network:
The controller maintains a complete map of the network, allowing for real-time decision-making and
optimization. This visibility enhances performance monitoring, security, and traffic engineering.
• Open Interfaces:
SDN uses open protocols like OpenFlow to connect controllers and devices. This reduces vendor
lock-in, fosters innovation, and allows integration of diverse devices from multiple manufacturers.

3) SDN Architecture (imp)

SDN Architecture contains the following three layers:
Application Layer:
The application layer is made up of applications which communicate with control layer controllers via certain
interfaces, called Northbound APIs.
The most used API is the REST (representation State Transfer) API for providing Northbound APIs.
Applications in SDN can be like Firewall, Load balancer etc.

Control Layer:
This is the middle layer in the SDN (Software-Defined Networking) architecture. It acts like the brain of the
network, making decisions about how traffic should flow. This layer is known as the control plane because it
controls and manages the behavior of network devices.
The SDN controller is usually installed on a physical server or a virtual machine. It talks to the data plane
(the part of the network that forwards data) using protocols like OpenFlow or other open APIs. These
protocols let the controller tell switches what to do with the data they receive.
The controller collects information about the network’s capacity (how much traffic it can handle) and current
demand (how much traffic is flowing). Using this information, it makes smart decisions to optimize the
network's performance.
SDN controllers also offer northbound APIs, which allow software applications to interact with the controller.
These APIs let network administrators and developers build new tools and services that run on top of the
network — such as monitoring tools, security features, or traffic analyzers.

Physical Layer:
includes equipment such as switches, also known as Data Plane, used in the network. They include
forwarding and switching of packets. Switches only perform the actions according to the controller.
The interface they use to communicate with a control layer is Southbound APls. OpenFlow Protocol is the
most common protocol used to provide southbound API.
4) Compare with diagram Modern Approach to computing and networking

Structural Comparison:
Modern approach to computing
In the modern approach to computing, there's a clear separation between the different layers of the system:
applications, operating systems (like Windows, Linux, or macOS), and the hardware (usually processors like
x86 or ARM).
These layers communicate using open interfaces, allowing flexibility and portability.
For example, the same application can be used across different devices because of common standards,
virtualization, and the ability of operating systems to work on multiple hardware platforms.

Modern approach to networking

On the other hand, the modern approach to networking using SDN (Software-Defined Networking) adopts
a similar layered model but applies it to network infrastructure. Here, applications refer to network control
programs, which sit on top of an SDN controller—the control plane—that manages the behavior of the
underlying switching hardware (the data plane).
This separation allows centralized and programmable control of network traffic, making the network more
dynamic and adaptable to changing demands

While both systems emphasize modularity and openness, the key difference is in purpose and function:
modern computing is focused on running software efficiently on hardware, while SDN is focused on
controlling how data moves through a network in a centralized and intelligent way.
 Functional Comparison:
(a) Traditional Network Architecture
• Each switch combines the control plane and data plane, meaning every switch makes its own routing
decisions.
• Packet flow and forwarding rules are managed locally on each device, which makes network-wide
updates slow and complex.
(b) SDN Approach
• The control plane is centralized in the SDN controller, which makes all routing decisions.
• Programmable switches handle only the data plane, forwarding packets based on instructions from
the controller.
• This structure allows for centralized control, real-time adjustments, and easier deployment of
network policies.

5) Openflow logical network device

An OpenFlow logical network device is a software-defined switch or router that follows rules defined by a central
controller. Instead of deciding on its own how to forward packets (like traditional network devices), it uses
instructions sent by a controller in a Software-Defined Networking (SDN) environment.
Key Components Explained:
1. SDN Controllers (Top)
o These are centralized control units that run the OpenFlow protocol.
o They manage all the OpenFlow switches below.
o Decide how traffic should flow and install flow rules into the switches.
2. OpenFlow Switches (Middle)
o Act as data plane devices that follow rules from the controller.
o They forward or drop packets based on the flow table.
Flow Table(s): Contains flow entries that include:
▪ Match fields (to identify packets)
▪ Instructions (what to do)
▪ Counters (stats)
▪ Timeouts and priority
o Communicate with the controller using OpenFlow protocol over TLS
3. End Systems (Bottom)
o These are users, servers, or any computing devices connected to the network.
o They send and receive packets which flow through the OpenFlow switches.
4. OpenFlow Protocol over TLS
o This is a secure communication channel between the controller and the switches.
o Allows the controller to:
▪ Add/update/remove flow entries
 ▪ Collect stats
▪ Receive event notifications

How It Works (Step-by-Step):

1. A packet arrives at the OpenFlow switch.

2. The switch checks its flow table for a matching rule.
o If a match is found: It applies the instruction (forward, drop, modify, etc.).
o If no match is found: It sends the packet info to the controller.
3. The controller decides what to do and sends a new rule back to the switch.
4. The switch updates its flow table and handles the packet.

6) Draw Diagram of OpenFlow Switch, Flow table fields in flow table structure

• The flow table represents the fundamental aspect of the logical switch architecture.
• Each packet entering a switch is loaded with one of the flow tables.

A flow table consists of flow entries.

Match Fields: These are used to identify specific packets based on characteristics like the ingress port,
packet headers, or metadata from previous tables. Only packets that match these fields are considered for
processing.
Priority: Determines the order of precedence when multiple flow entries match a packet. The entry with the
highest priority is chosen.
Counters: These keep track of statistics such as how many packets or bytes matched a particular flow entry,
helping in monitoring and analytics.
Instructions: These specify what actions to take on matching packets, such as modifying headers, sending
packets to a specific port, or directing them through further processing stages.
Timeouts: These define how long a flow entry remains active. It can be based on idle time (no matching
packets for a period) or a hard timeout (fixed maximum duration), after which the entry is removed.
Cookie: A unique identifier set by the controller. It is not used for packet processing but helps the controller
manage and filter flow statistics, updates, or deletions related to specific flows.
7) Explain OpenDayLight Architecture
OpenDaylight is a Linux Platform hosting open source SDN controller / framework. It is currently one of the most
popular SDN controllers (open source). One of the protocols for the southbound interface is OpenFlow.

Network applications, orchestration, and services: The top layer includes business and network logic applications
which control and monitor the computation of the network.
APIs in OpenDaylight:
OpenDaylight provides a set of standard controller functions, called APIs. These help applications communicate
with the controller.
It supports Open Service Gateway Initiative (OSGi), a framework that allows apps to run inside the controller’s
system.
For apps running outside the controller (even on different systems), OpenDaylight uses REST APIs, which work
over the web.

Controller Functions & Services:

• Network Service Functions – like path computation, topology discovery, etc.
• Platform Services – support functions like data storage, message passing, etc.
• Extensions – extra features or custom modules added to enhance the controller.

Service abstraction layer (SAL):

The Service Abstraction Layer (SAL) is a key part of OpenDaylight's design. It acts like a middleman between
services (producers) and the apps that use them (consumers).
Here’s what SAL does:
• Keeps a list (registry) of all the services available in different modules.
• Connects the right service to the right app when it's needed.
• Makes sure requests from apps go to the correct service, using standard APIs.

Southbound interfaces and protocols:

The southbound interface connects OpenDaylight to network devices using different protocols like:
OpenFlow (1.0 & 1.3)
• BGP-LS Border Gateway Protocol - Link State
• LISP Locator/ID Separation Protocol
• SNMP, Simple Network Management Protocol
 These protocols work as plugins and are linked to the Service Abstraction Layer (SAL).
SAL uses these plugins to figure out how to carry out the service requested by apps, no matter which protocol or
device is being used in the network.

Data Plane Elements.

• This includes the actual network hardware or virtual switches (like routers, switches, and firewalls).
• These are the devices OpenDaylight controls using the southbound protocols.

8) Explain REpresentational State Transfer (REST)

• REST is REpresentational State Transfer, and API stands for Application Program Interface. REST is an
architectural software style that sets the rules to create web services.
• The REST-architectural type web services are known as the RESTful web services. It allows for systems to use
a standardized, default collection of standards to access and manipulate web resources.
• REST-based systems communicate via the Hypertext Transfer Protocol (HTTP).
A Restful system consists of a:
• Client who requests for the resources.
• server who has the resources.
It is necessary to construct a REST API in line with industry standards that will make development easier and
customer adoption easier.

REST Constraints:
• Uniform Interface
• Stateless
• Cacheable
• Client-Server
• Layered System
• Code on Demand

Uniform Interface: This is the main restriction between REST API and Non-Rest API, indicating that a uniform
interaction with a server should be defined regardless of device or application type like uniform interface
(website, mobile app).
Stateless: Stateless means the server doesn’t remember past requests—so every time the client (like a browser
or app) sends a request, it must include all the needed information in the URL, headers, or query parameters,
which can increase bandwidth usage.
Cacheable: means the server tells the client if it can save the response and for how long. If allowed, the client
can reuse the saved data later without asking the server again, which makes things faster—but sometimes the
data might be outdated.
Client-Server:
• Client: Requests resources and handles the user interface but doesn’t manage data or business logic.
• Server: Stores data and processes business logic, but doesn’t deal with the client’s UI.
Layered System:
• The application is divided into multiple layers, each unaware of the others beyond the immediate one.
• Between the client and server, there can be several intermediate layers (servers).
• These intermediate layers help with load balancing and offer shared caching to improve performance
and increase system availability.
Code on Demand: This is an option. This also means that servers can provide the client with executable code. For
example, compiled components such as Java applets, client-side scripts such as JavaScript can include code on
demand.
9) Explain Traffic Engineering
Traffic Engineering refers to the process of optimizing the flow of data across a network to ensure efficient use of
resources and to improve performance.
In the context of applications or networks, it involves techniques like:

• Load Balancing
o Definition: Distributing incoming traffic evenly across multiple servers or network resources to
prevent overloading any single resource.
o How it works: Traffic can be distributed based on various algorithms, such as round-robin, least
connections, or weighted balancing, ensuring no server is overwhelmed, which leads to better
performance and availability.

• Quality of Service (QoS):

o Definition: Prioritizing certain types of network traffic over others to ensure that high-priority traffic
(e.g., voice or video) gets the necessary resources for a smooth experience.
o How it works: Network devices (like routers and switches) can tag traffic with different levels of
priority. QoS mechanisms ensure high-priority traffic is transmitted with minimal delay, while lower-
priority traffic is delayed or dropped if necessary.

• Routing Optimization:
o Definition: Choosing the most efficient route for data packets across a network to avoid congestion,
reduce delays, and improve overall performance.
o How it works: Routing algorithms, such as Shortest Path First (SPF), Equal-Cost Multi-Path (ECMP),
or Traffic Engineering Database (TED), determine the best path for traffic based on factors like
bandwidth, latency, and current network conditions.

• Network Monitoring:
o Definition: Continuously monitoring network traffic to identify issues like congestion, downtime, or
underutilized resources.
o How it works: Tools and software can analyze traffic patterns, detect anomalies, and identify
performance bottlenecks.

• Redundancy and Failover:

o Definition: Providing backup routes and resources in case of network failure.
o How it works: Networks are often designed with multiple paths between source and destination. In
case one path fails, traffic is automatically rerouted to an alternative path, ensuring continuous
service.

10) Big Data Network Ecosystem

Big data refers to everything that enables an organization to create, manipulate, and manage very large data sets
(measured in terabytes, petabytes, exabytes, and so on) and the facilities in which these are stored.
Top-Level Data Sources (Edge Layer)
1. Sensors and Other IoT Input Devices
o These are smart devices (like cameras, temperature sensors, etc.) that collect real-time data.
o In modern networking, they are part of the Edge Computing paradigm, where some processing
happens close to where data is generated to reduce latency.
2. Other Enterprise Data Sources
o These could be internal databases, software systems, or legacy infrastructure.
o They connect to the broader ecosystem through secure network channels (e.g., VPN, dedicated
links).
3. Content Delivery Network (CDN)
o CDNs store and deliver content (like videos or software updates) close to the user to reduce load
time and bandwidth usage.
o In big data, CDNs help distribute large-scale content efficiently across regions.

Internet and Other Networks (Core Layer)

• The Internet and Other Networks (Core Layer) acts as the secure, scalable backbone that connects data
sources to the enterprise environment using cloud infrastructure, hybrid networking, SD-WAN, and high-
speed data transfer.
Enterprise Network (Data Center/Cloud Layer)
This layer hosts the infrastructure that manages, processes, and stores data:
1. Data Management Blade Servers
o High-performance computing nodes designed for big data processing.
o These support frameworks like Hadoop, Spark, or cloud-based data platforms.
2. Data Warehouse
o Centralized storage for structured big data used for analytics and reporting.
 o Often cloud-hosted (e.g., Snowflake, Google BigQuery) or on-premise.
3. Network Management Server
o Monitors and controls the health, traffic, and security of the network.
o Uses tools for traffic engineering, network telemetry, and AI-based monitoring.
4. Workstations
o These are the user interfaces where analysts, engineers, or developers interact with data.
o Typically connected via secure enterprise LANs or remote access tools.

11) Cloud Network Model

1.Cloud Side (Left)

Data Centers (1, 2, and 3):
o Each data center contains multiple databases (DB1, DB2, DB3).
o These are physical or virtual servers used to store and manage big data in the cloud.
o These are hosted on cloud infrastructure provided by vendors like AWS, Azure, or GCP.
Virtual Network:
o A secure, software-defined network connecting the data centers.
o Ensures that cloud resources can talk to each other and route traffic internally.

2. Gateway (Center)
o Acts as a secure bridge between cloud resources and the on-premises infrastructure.
o Ensures safe and reliable data exchange using encryption and firewall rules.

3.On-Premises Network (Right)

• Contains routers, switches, and virtual machines (VMs) organized into virtual networks.
• These internal systems can access cloud-hosted databases and services through the gateway.

4.Cloud Services Layers (Integrated into the Model)

IaaS (Infrastructure as a Service)

o Provides virtual machines, storage, and networks over the cloud.
o Example: AWS EC2, which can host the databases shown in the cloud data centers.
PaaS (Platform as a Service)
o Offers a development environment with tools to build and run applications without managing
infrastructure.
o Example: Heroku, which could be used to develop apps that pull data from these cloud databases.
 SaaS (Software as a Service)
o Fully functional software applications accessed through the internet.
o Example: Google Workspace, which could be used by employees on the on-premises network to
collaborate using cloud data.

12) High-Level Architecture ITU-T MODEL

SDN functional architecture is based on the SDN framework [ITU-T] Model.

The control layer provides the means, as directed by the application layer, to dynamically manage network
resources. The control layer can be seen as providing the following sublayers:
Application support:
The application support function provides application control interface to SDN applications to access network
information details and application-specific behavior of the program.
Orchestration:
The orchestration function provides network infrastructure control and management, such as management
of physical and virtual network topologies, network elements and traffic. It integrates with multi-layer
management features to manage SDN applications such as user management, service advancement and
distribution.
Abstraction:
The Abstraction Function interacts with network resources and gives an overview of network resources,
including network capacity and features that support management and orchestration of physical and virtual
network resource.
Resource layer:
The resource layer is used to transport and process data packets by the network elements based on the
decisions taken by the SDN control layer and distributed through a resource control interface to the resource
layer.
Control support:
The control support function interfaces with and maintains the SDN control layer programming through
resource-control interfaces
Data transport and processing:
Data transfer and processing function includes data forwarding and data routing functionality. The data
forwarding function manages the incoming data flows to forward them along the data transfer routes which
were calculated and computed according to the SDN applications' requirements. The data forwarding
functionality is managed by the SDN control layer in order to reduce the data transfer functionality in the
resource layer.