COMPUTER NETWORKS

23DC2008

- MODULE 3
 Network Layer - Module: 3 -7 Hours

•Forwarding and Routing, Router Architecture, IPv4

addressing, IPv6 addressing, Generalized Forwarding
and SDN, Routing Algorithms – Link State Routing
Algorithm, Distance Vector Routing Algorithm,
OSPF, BGP, ICMP, Network Management and
SNMP, NETCONF/YANG.
 Link Layer - Module: 4 - 8 Hours

•Link Layer Services, Error Detection and

Correction Techniques, Multiple Access Links
and Protocols, Switched Local Area Networks,
Link Virtualization, Data Center Networking.
Security in Computer Networks - Module: 5 - 8 Hours

•Principles of Cryptography, Message Integrity

and Digital Signatures, End Point Authentication,
Securing Email, Securing TCP Connections,
Network Layer Security, Operational Security.
 Wireless and Mobile Networks - Module: 6 - 8 Hours

•Wireless Links and Network Characteristics,

WiFi 802.11 Wireless LANs, Cellular Networks:
4G and 5G, Securing Wireless LANs and 4G/5G
Cellular Networks, Mobility Management.
Network Layer – (I) Forwarding &(II) Routing
• A Router is involved in transferring a packet from an incoming link to
an outgoing link.
• FORWARDING & ROUTING
• To move packets from a sending host to receiving host
• Two important network-layer functions can be identified:
• FORWARDING &
• ROUTING
 1) Forwarding
• Packet arriving at the Routers’ input must be moved to appropriate
link.

• Sometimes a packet may also be prevented from exited from a

Router.
•
 2) Routing: Redirecting of packet to the
desired link
• N/W layer must determine the route or path taken by packets while
flowing from sender to receiver.
• Routing algorithms may be centralized or decentralized
• Routers work with Algorithms called Routing algorithms finding paths
through which packets flow.
• Every Router has a forwarding Table.
• Routing refers to network-wide process that determines the
end-to-end paths to that packets take from source to destination.
ROUTING ALGORITHMS & FORWARD TABLES
 Packet Switch
• Interfaces exist in Input link as well as Output link.

• General packet switching N/W device called packet switch for

transferring a packet from I/P interface to O/P interface.

• Packet Switch have lots of similarity with Routers

• Apart from Forwarding and Routing, N/W have CONNECTION SETUP.

CONTROL PLANE with Forwarding Table
- Remote Controller
 Network Service Model
-Services (Best – Effort) offered by N/W Model
• 1) Guaranteed delivery

• 2) Guaranteed delivery with bounded delay

• 3) In-order packet delivery

• 4) Guaranteed minimal bandwidth

• 5) Security
 Router - Overview
•Routing or Redirection of Data units is done by this H/W
networking device within the channel to the desired
direction
ROUTER ARCHITECTURE
 Input ports, Switching Fabric
• I/P Port:
• Physical Layer element doing key functions to interoperate with Link
Layer, at the side of Input link.
• This port takes the Control Packets to Routing Processor
• The number of ports can support hundreds to 10 Gpbs.
• Switching Fabric
• Input ports are connected to output ports through Switch Fabrics
• Interlinking of Routers are an example of such a Fabric.
 Output ports, Routing Processor

•Output Port:
• The packet traffic is collected from Switch Fabric by O/P port.
• This port will take the packets to the output link.
• Links can be bidirectional, so the porting will be paired with I/P port
too.
•Routing Processor
•This is for executing the routing protocols, maintaining
routing information, forwarding tables and N/W
management functions.
Processing of Input Port
 Switching & I/P ports
- Memory-based (e.g. Cisco - 8500 series)
 -Bus-based Switch
- (e.g. Cisco 5600 switch – 32 Gbps )
- Crossbar-based Switch
- e.g. Cisco 5600
Processing by output Port
Input Queuing
Packet & HOL Blocking
- Head-of-Line
Output Queueing
 Packet Scheduling
• Queued packets are transmitted over outgoing link in a specified
order
• Waiting customers are served in a schedule:
• 1) FCFS
• 2) Round-Robin
• 3) Priority-based
First-in-First-Out (FIFO) –Link scheduling discipline
- also does Buffer management in Queue
Priority Queuing
HIGH-PRIORITY QUEUE
 WEIGHTED FAIR QUEUING(WFQ)
- Arriving packets classified, queued fairly
 IPv4 Addressing
• IP addresses are Architecture-Independent, globally
network-oriented
 IPv4 Datagram Format –Syntax & Semantics

• Version: number of Version – 4 bits

• Header length: 20 B header
• Type of service: TOS
• e.g. IP telephony for real-time datagrams
• FTP for non-real time datagrams
• Datagram length
• Total length = (Header plus Data)
• 16-bits-maximum size of the IP datagram is 65,535 bytes.
 Fields of IPV4 format
• Identifier, Flags, Fragmentation offset
• This is about Identifier, Flags & knowing Fragments ( 3 Fields)
• Time-to-live (TTL)
• Duration of time to start and reach a packet in N/W. Life a
packet inside N/W till it is discarded.
• Protocol:
• Specifies the Transport-Layer Protocol
• Header checksum: For error handling purposes.
• Routers can be helped by this field. It detects bit in a
received IP datagram.
• Source & destination IP addresses. Starting and Ending point of an IP
datagram.
….
• Options :
• Extendable IP header options and details.
• Data (payload): Data Field / Content to be transferred to other point.
•
 IP FRAGMENTATION & RE-ASSEMBLY
- Large datagram divided and re-joined again
 IPV4 ADDRESSING
• When a Host IP needs to dispatch datagram , it used link inside the
N/W
• An Interface is the boundary b/w the HOST & link which enables
Datagram that needs some pointing means to reach.
• A router has multiple Interfaces (each associated with IP address) for
Sending/ Receiving Data Units.
32
• Each 32 bits long thus 2 possibilities.
• 4 billion possible IP addresses.
INTERFACE ADDRESSES & SUBNETS
 ADDRESS SUBNETS
- Subset of isolated hosts
 Classless Interdomain Routing - CIDR
• CIDR defines and specifies the policy of IP addressing and subnet
addressing
 DHCP-Client-server protocol
• For obtaining a Host Address
• Configuring IP address in Router
 IPv6 - Datagram Format
• For a larger IP address space IPv6 has to be formed with 128-bit length
• Expanded addressing capabilities
• IP addresses size increased from 32 to 128 bits.
• Anycast address also possible in IPV6
• A streamlined 40-byte header
• 40-Byte fixed length header
• Faster processing of IPv6 datagram
• Flow labelling – IPV6 gives a specific labelling of packets for Flow to End.
Format of IPV6 - Datagram
 Fields in IPV6 Datagram Structure
• Version: Identifies 4-bit the IP version
• Traffic class: 8-bit traffic class field
• Flow label: 20 B field for identification of datagram flow.
• Next header : Identifies to which the data field will be sent.
• Hop limit : Hop means each forwarding in link. The permitted limit is
this
• Source and Destination addresses : Formats of addresses
• Data: Payload Portion
 Fields devoid in IPV6 Datagram Structure
•Fragmentation/Reassembly
• These are good for IPV4, NOT suited to IPV6.
•Header checksum:
• For Transport layer to performing CheckSumming
•Options:
• Any additional Information.
Transitioning from IPv4 to IPv6
example: Tunnelling
 Generalized Forwarding
• Includes Forwarding with Switches, NAT Boxes, specialized H/W
• Unified Approach, modern, elegant, and integrated
• Generalized Forwarding includes the study of:
• Match,
• Action,
• Destination-based forwarding,
• Load balancing,
• Blocking/Dropping a packet
 Overview of Generalized Forwarding
-Decision making
• N/W layer process of packet transfer via ROUTERS
• Packets are for GENERALIZED FORWARDING in order to directed to
destination
• Especially good for UNICAST
• Router uses:
• 1) Tables for routing
• 2) Decisions for Forwarding
 SDN- Overview
• Software-defined networking is a standard approach for a N/W
architecture in order to get controlled or programmed using s/w
applications.

• This SDN is regardless of any Technology of N/W(s)

• API and S/W-based Controllers are extensively used to traffic control

and can give rise to Virtual Networks.
 VIRTUAL CIRCUITS - VC
• These are circuits using connections at N/W layer.
• Routers work with link interfaces.
• A link connects itself with other nodes forming PATH.
• VC consists of :
• 1) A path b/w SRC and DESTN
• 2) VC number
• 3) Forwarding table entries
NETWORK ADDRESS TRANSLATION
NAT – WORKING SCHEME
 SDN - Significance and Working
SIGNIFICANCE
• 1. Increased control with greater speed and flexibility
• 2. Infrastructure of N/W that is customizable
• 3. Security that is Robust
• WORKING of SDN :
• 1. Applications
• 2. Controllers &
• 3. Networking Devices
 Network Layer: Control Plane
by NL Control logic
• A datagram is forwarded along End-to-End through Routers from SRC host to
DEST host
LOGICALLY CENTRALIZED CONTROL
 Routing Algorithms
• Algorithms to Determine the best path / route from Sender to
Receiver through N/W of Routers with the least cost compute.

• Graph G=(N, E) ; N nodes and E edges where G is analyzed in NL

routing
 ABSTRACT GRAPH MODEL
- Neighbor Nodes, Costs, Edges, Least Cost computation, Forwarding decisions
 ROUTING ALGORITHMS
• All least cost path may not necessarily be SHORTEST PATH.
• Routing algorithms may be CENTRALIZED - SRC and DESTN has paths
having costs and take least cost.
• -Gets connectivity between all nodes and all link costs as inputs

• Cost is computed as the least-cost path (LCP)

• LINK-STATE (LS) ALGORITHMS
• LS algorithms involve Global state Information and LS algorithm must be
aware of cost of each link.
• DECENTRALIZED ALGORITHM
• Here Router calculates Least-cost path in distributive & repetitive manner.
 DECENTRALIZED ROUTER ALGORITHM
• Routers calculate least-cost path by ITERATIVE, DISTRIBUTIVE manner

• Contrasted to Centralized arrangement to Route packets to paths

• The algorithmic computing is NOT performed by server

DETAILS IN A ROUTING TABLE
Popular Protocols - ROUTING
 Link-State (LS) Routing
- Link, Destination, Forwarding Table
 LS algorithm – How it works?
• The costs of link is known and also topology as the INPUT to
algorithm
• Adjacent NODES are connected as LINKS whereas entire way or travel
is referred as PATH
• LS algorithm closely follows the Graph Theory Algorithms like:
• Dijkshtra’s Algorithm &
• Prim’s Algorithm.
• Each NODE can run LS algorithm
• Same set of least-cost paths as every other node.
Terms in LS Algorithm
LS Run – Distance of each Node from u.
Least cost of all links
 Distance-Vector (DV) Routing Algorithm
- Unlike LS algorithm which uses global data, DV routing algorithm uses:
- DISTRIBUTED,
&
- ASYNCHRONOUS,
- ITERATIVE.
- From every neighbor of a NODE, data is arriving to it -> performs
calculations -> then re-distributed to OTHER NODES
- The process ITERATES until no information-exchange occurs
- DV algorithm asynchronous , so that all NODES need not operate at same
step and time.
 Features and Uses of DV algorithm
• DV is DECENTRALIZED VECTOR
• DV algorithms are used in protocols like:
• - Open Shortest Path First (OSPF)
• - Border Gateway Protocol (BGP)
• - Internet Control Message Protocol (ICMP)
• -Simple Network Management
Protocol (SNMP)
• - Network Configuration Protocol (NCP)
Working Of Distance Vector Algorithm
 LS versus DV Routing
• In DV Routing, each NODE “talks” to adjacent ( immediate )
• But LS takes / needs global information for steps
• These two Routing differ:

1) Message Complexity

2) Speed of Convergence

3) Robustness
 OSPF:Intra-AS Routing in the Internet
-AS – Autonomous Systems
• Dynamic IP protocol for Routing tasks for finding most efficient paths in
Large-scale N/W
• OSPF – Open Shortest Path First to calculate the best path
• OSPF operates under the control of same administration
• OSPF is link-state protocol containing link-state information
• Path has various Factors/Metrics for OSPF calculation like link’s :
• 1) Bandwidth
• 2) Cost
• 3) Delay
 OSPF-HOW IT WORKS?
• All other Routers (except that broadcasting Router) avails link-state
information receives
• This link-state information and message transfer are implemented as
functionalities by OSPF protocol
• BENEFITS:
• 1) SECURITY
• 2) MULTIPLE SAME-COST PATHS
• 3) INTEGRATED SUPPORT FOR UNICAST AND MULTICAST ROUTING
• 4) SUPPORT FOR HIERARCHY WITHIN A SINGLE AUTONOMOUS SYSTEMS
 BORDER GATEWAY PROTOCOL (BGP)
• This protocol can connect any multiple autonomous system (AS)
which has atleast one Router in it using an arbitrary topology.

• Main function of BGP is to exchange information with other BGP

systems.

• Countries, Organizations and ISPs can be interconnected through BGP

with Routing information (Quick and Seamless communication)
- IANA (Internet Assigned Numbers Authority) is the number in AS for
identifying BGP protocol- N/W.
BGP – BORDER GATEWAY PROTOCOL
- Architecture
 Factors affecting the decision of finding best route
from SRC -> DESTN
• 1) Data Transfer Cost
• - Should be minimal

• 2) Network Congestion
• - least one

• 3) Geographical Location
• - well connected in sites
 Routing Among the ISPs
• Packets are CIDR-ized prefixed , NOT just forwarded to outbound link
• A Router in BGP-N/W offers information like:
• 1) Neighboring Autonomous S/M will obtain prefix
reachability data
• 2) Among the routes to the prefix, find the best route.
• This Router informs Prefix information to BGP connected NODES and
this includes prefix with BGP attributes.
• ROUTE is the term representing PREFIX along attributes
BGP N/W connectivity – 3 AS having subnets
Internet Control Message Protocol – ICMP
- considered a part of IP
- ICMP messages are inside IP datagram
•Runs in N/W Layer for information layer communication
used (e.g. Error Reporting ) by Hosts and Routers.
•ICMP issues messages like “Host Not Reachable” in HTTP
session.
•The Payload field of IP structure contains ICMP message
•Example of an ICMP message is ping type 8 code 0
•Various types of ICMP messages are available for
Connectivity
 ICMP message in Traceroute program (RFC 4443)

• Traceroute is a program implemented using ICMP message to track

and know the details (name, address) of Intermediate Routers b/w
SRC and DESTN

• First, the SRC traceroute sends a series of IP datagrams to the

destination where each datagram carries UDP segment.

• This communication will be associated with WARNING messages, if

needed inside Payload field of UDP.
 Network Management and SNMP
- Simple N/W Management Protocol
• N/W managing is the most engineering effort of all categories of N/W
everywhere since many complex, interactive, sophisticated
technology involvements makes it highly effortful.
 Framework for network management
• List of key factors for managing N/W are:
•A) Managing Server
•B) Device that is managed
•C) Management Information Base (MIB)
•D) Agent for N/W managing
•E) N/W management Protocol
NETWORK MANAGEMENT FRAMEWORK
- MIB data, Agent, Device, Server, SNMP
A) Managing Server
• Centralized Part of managing stations
• Does activities like:
• 1) Process N/W information,
• 2) Collects them , Analyze and display these information.
B) Device that is managed
• Managed Piece of H/W + S/W residing on the managed N/W

• Example:
• HOST,
• ROUTER,
• BRIDGE,
• HUB,
• PRINTER,
• MODEM.
C) Management Information Base (MIB)
• Managed Information has pieces of information collected into MIB

• Example : Details on
• Activity, Budget, Productivity and the like.
D) Agent for N/W managing
• Network Management Agent (NMA):
• -They reside in each N/W device

• - NMA are action programs running in N/W managed devices

• - Takes local actions that communicates with managing entity

E) N/W management Protocol
• N/W management protocol is a part of architecture running b/w
managing entity and managed device

• Agents can query and get the status of managed device.

 INTERNET STANDARD MANAGEMENT FRAMEWORK

• This is a standard Framework which follows from SIMLPE GATEWAY

MONTIROING PROTOCOL (SGMP)
• - Designs, Deploys, implement SNMP
• 4 parts addressed by this Framework are:
• 1) Definition of n/w management objects (MIB)
• 2) Data Definition Language (DDL): information defining data
types,(integer, Object ID, IP address) an object model, and writing
rules of MIB objects
• 3) Protocol , SNMP
• 4) Security and administration capabilities
 NETCONF/YANG.

• Standard way of Configuring of a N/W device in large-scale Network.

• NETCONF is a protocol that can programmatically configure a device
within N/W
• IT requires a language for modelling all these tasks of configuration
• YANG is the Modelling language serving this purpose
• NETCONF commands operates via CLI
• YANG can aim on our desired state or configuration.
• Cisco IOS XE 16.X Platforms are available.
-
•-
Yang Explorer UI screen