UNIT-III

Network Layer Design Issues

1. Store-and-forward packet switching
2. Services provided to transport layer
3. Implementation of connectionless service
4. Implementation of connection-oriented service
5. Comparison of virtual-circuit and datagram networks

1 Store-and-forward packet switching

A host with a packet to send transmits it to the nearest router, either on its own LAN or over a
point-to-point link to the ISP. The packet is stored there until it has fully arrived and the link
has finished its processing by verifying the checksum. Then it is forwarded to the next router
along the path until it reaches the destination host, where it is delivered. This mechanism is
store-and-forward packet switching.

2 Services provided to transport layer

The network layer provides services to the transport layer at the network layer/transport layer
interface. The services need to be carefully designed with the following goals in mind:
1. Services independent of router technology.
2. Transport layer shielded from number, type, topology of routers.
3. Network addresses available to transport layer use uniform numbering plan
– even across LANs and WANs

3 Implementation of connectionless service

If connectionless service is offered, packets are injected into the network individually and
routed independently of each other. No advance setup is needed. In this context, the packets
are frequently called datagrams (in analogy with telegrams) and the network is called a
datagram network.

A’s table (initially) A’s table (later) C’s Table E’s Table

Let us assume for this example that the message is four times longer than the maximum
packet size, so the network layer has to break it into four packets, 1, 2, 3, and 4, and send each
of them in turn to router A.
Every router has an internal table telling it where to send packets for each of the possible
destinations. Each table entry is a pair(destination and the outgoing line). Only directly
connected lines can be used.
A’s initial routing table is shown in the figure under the label ‘‘initially.’’
At A, packets 1, 2, and 3 are stored briefly, having arrived on the incoming link. Then each
packet is forwarded according to A’s table, onto the outgoing link to C within a new frame.
Packet 1 is then forwarded to E and then to F.
However, something different happens to packet 4. When it gets to A it is sent to router B,
even though it is also destined for F. For some reason (traffic jam along ACE path), A decided
to send packet 4 via a different route than that of the first three packets. Router A updated its
routing table, as shown under the label ‘‘later.’’
The algorithm that manages the tables and makes the routing decisions is called the routing
algorithm.
4 Implementation of connection-oriented service

A’s table C’s Table E’s Table

If connection-oriented service is used, a path from the source router all the way to the
destination router must be established before any data packets can be sent. This connection is
called a VC (virtual circuit), and the network is called a virtual-circuit network

When a connection is established, a route from the source machine to the destination
machine is chosen as part of the connection setup and stored in tables inside the routers. That
route is used for all traffic flowing over the connection, exactly the same way that the
telephone system works. When the connection is released, the virtual circuit is also
terminated. With connection-oriented service, each packet carries an identifier telling which
virtual circuit it belongs to.

As an example, consider the situation shown in Figure. Here, host H1 has established
connection 1 with host H2. This connection is remembered as the first entry in each of the
routing tables. The first line of A’s table says that if a packet bearing connection identifier 1
comes in from H1, it is to be sent to router C and given connection identifier 1. Similarly, the
first entry at C routes the packet to E, also with connection identifier 1.
Now let us consider what happens if H3 also wants to establish a connection to H2. It chooses
connection identifier 1 (because it is initiating the connection and this is its only connection)
and tells the network to establish the virtual circuit.
This leads to the second row in the tables. Note that we have a conflict here because although
A can easily distinguish connection 1 packets from H1 from connection 1 packets from H3, C
cannot do this. For this reason, A assigns a different connection identifier to the outgoing
traffic for the second connection. Avoiding conflicts of this kind is why routers need the ability
to replace connection identifiers in outgoing packets.
In some contexts, this process is called label switching. An example of a connection-oriented
network service is MPLS (Multi Protocol Label Switching).

5 Comparison of virtual-circuit and datagram networks

Routing Algorithms
The main function of NL (Network Layer) is routing packets from the source machine to the
destination machine.
There are two processes inside router:
a) One of them handles each packet as it arrives, looking up the outgoing line to use for it in
the routing table. This process is forwarding.
b) The other process is responsible for filling in and updating the routing tables. That is where
the routing algorithm comes into play. This process is routing.

Regardless of whether routes are chosen independently for each packet or only when new
connections are established, certain properties are desirable in a routing algorithm
correctness, simplicity, robustness, stability, fairness, optimality
 Routing algorithms can be grouped into two major classes:
1) nonadaptive (Static Routing)
2) adaptive. (Dynamic Routing)

Nonadaptive algorithm do not base their routing decisions on measurements or estimates of

the current traffic and topology. Instead, the choice of the route to use to get from I to J is
computed in advance, off line, and downloaded to the routers when the network is booted.
This procedure is sometimes called static routing.

Adaptive algorithm, in contrast, change their routing decisions to reflect changes in the
topology, and usually the traffic as well.
Adaptive algorithms differ in
1) Where they get their information (e.g., locally, from adjacent routers, or from all routers),
2) When they change the routes (e.g., every ∆T sec, when the load changes or when the
topology changes), and
3) What metric is used for optimization (e.g., distance, number of hops, or estimated transit
time).
This procedure is called dynamic routing

Different Routing Algorithms

• Optimality principle
• Shortest path algorithm
• Flooding
• Distance vector routing
• Link state routing
• Hierarchical Routing

The Optimality Principle

One can make a general statement about optimal routes without regard to network topology
or traffic. This statement is known as the optimality principle.
It states that if router J is on the optimal path from router I to router K, then the optimal path
from J to K also falls along the same
As a direct consequence of the optimality principle, we can see that the set of optimal routes
from all sources to a given destination form a tree rooted at the destination. Such a tree is
called a sink tree. The goal of all routing algorithms is to discover and use the sink trees for all
routers
 (a) A network. (b) A sink tree for router B.

Shortest Path Routing (Dijkstra’s)

The idea is to build a graph of the subnet, with each node of the graph representing a router
and each arc of the graph representing a communication line or link.
To choose a route between a given pair of routers, the algorithm just finds the shortest path
between them on the graph
1. Start with the local node (router) as the root of the tree. Assign a cost of 0 to this node and
make it the first permanent node.
2. Examine each neighbor of the node that was the last permanent node.
3. Assign a cumulative cost to each node and make it tentative
4. Among the list of tentative nodes
a. Find the node with the smallest cost and make it Permanent
b. If a node can be reached from more than one route then select the route with the
shortest cumulative cost.
5. Repeat steps 2 to 4 until every node becomes permanent
Flooding
• Another static algorithm is flooding, in which every incoming packet is sent out on every
outgoing line except the one it arrived on.
• Flooding obviously generates vast numbers of duplicate packets, in fact, an infinite number
unless some measures are taken to damp the process.
• One such measure is to have a hop counter contained in the header of each packet, which
is decremented at each hop, with the packet being discarded when the counter reaches
zero. Ideally, the hop counter should be initialized to the length of the path from source to
destination.
• A variation of flooding that is slightly more practical is selective flooding. In this algorithm
the routers do not send every incoming packet out on every line, only on those lines that
are going approximately in the right direction.
• Flooding is not practical in most applications.

Intra- and Inter domain Routing

An autonomous system (AS) is a group of networks and routers under the authority of a single
administration.
Routing inside an autonomous system is referred to as intra domain routing. (DISTANCE
VECTOR, LINK STATE)
Routing between autonomous systems is referred to as inter domain routing. (PATH VECTOR)
Each autonomous system can choose one or more intra domain routing protocols to handle
routing inside the autonomous system. However, only one inter domain routing protocol
handles routing between autonomous systems.

Distance Vector Routing

In distance vector routing, the least-cost route between any two nodes is the route with
minimum distance. In this protocol, as the name implies, each node maintains a vector (table)
of minimum distances to every node.

Mainly 3 things in this

Initialization
Sharing
Updating

Initialization
Each node can know only the distance between itself and its immediate neighbors, those
directly connected to it. So for the moment, we assume that each node can send a message to
the immediate neighbors and find the distance between itself and these neighbors. Below fig
shows the initial tables for each node. The distance for any entry that is not a neighbor is
marked as infinite (unreachable).
Initialization of tables in distance vector routing

Sharing
The whole idea of distance vector routing is the sharing of information between neighbors.
Although node A does not know about node E, node C does. So if node C shares its routing
table with A, node A can also know how to reach node E. On the other hand, node C does not
know how to reach node D, but node A does. If node A shares its routing table with node C,
node C also knows how to reach node D. In other words, nodes A and C, as immediate
neighbors, can improve their routing tables if they help each other.
NOTE: In distance vector routing, each node shares its routing table with its immediate
neighbors periodically and when there is a change

Updating
When a node receives a two-column table from a neighbor, it needs to update its routing
table. Updating takes three steps:
1. The receiving node needs to add the cost between itself and the sending node to each value
in the second column. (x+y)
2. If the receiving node uses information from any row. The sending node is the next node in
the route.
3. The receiving node needs to compare each row of its old table with the corresponding row
of the modified version of the received table.
a. If the next-node entry is different, the receiving node chooses the row with the
smaller cost. If there is a tie, the old one is kept.
b. If the next-node entry is the same, the receiving node chooses the new row.
For example, suppose node C has previously advertised a route to node X with distance 3.
Suppose that now there is no path between C and X; node C now advertises this route with a
distance of infinity. Node A must not ignore this value even though its old entry is smaller. The
old route does not exist anymore. The new route has a distance of infinity.

Updating in distance vector routing

Final Diagram
When to Share
The question now is, When does a node send its partial routing table (only two columns) to all
its immediate neighbors? The table is sent both periodically and when there is a change in the
table.
Periodic Update A node sends its routing table, normally every 30 s, in a periodic update. The
period depends on the protocol that is using distance vector routing.
Triggered Update A node sends its two-column routing table to its neighbors anytime there is
a change in its routing table. This is called a triggered update. The change can result from the
following.
1. A node receives a table from a neighbor, resulting in changes in its own table after updating.
2. A node detects some failure in the neighboring links which results in a distance change to
infinity.

Two-node instability

Three-node instability

SOLUTIONS FOR INSTABILITY

1. Defining Infinity: redefine infinity to a smaller number, such as 100. For our previous
scenario, the system will be stable in less than 20 updates. As a matter of fact, most
implementations of the distance vector protocol define the distance between each node to
 be 1 and define 16 as infinity. However, this means that the distance vector routing cannot
be used in large systems. The size of the network, in each direction, cannot exceed 15 hops.

2. Split Horizon: In this strategy, instead of flooding the table through each interface, each
node sends only part of its table through each interface. If, according to its table, node B
thinks that the optimum route to reach X is via A, it does not need to advertise this piece of
information to A; the information has come from A (A already knows). Taking information
from node A, modifying it, and sending it back to node A creates the confusion. In our
scenario, node B eliminates the last line of its routing table before it sends it to A. In this
case, node A keeps the value of infinity as the distance to X. Later when node A sends its
routing table to B, node B also corrects its routing table. The system becomes stable after
the first update: both node A and B know that X is not reachable.

3. Split Horizon and Poison Reverse Using the split horizon strategy has one drawback.
Normally, the distance vector protocol uses a timer, and if there is no news about a route,
the node deletes the route from its table. When node B in the previous scenario eliminates
the route to X from its advertisement to A, node A cannot guess that this is due to the split
horizon strategy (the source of information was A) or because B has not received any news
about X recently. The split horizon strategy can be combined with the poison reverse
strategy. Node B can still advertise the value for X, but if the source of information is A, it
can replace the distance with infinity as a warning: "Do not use this value; what I know
about this route comes from you."

The Count-to-Infinity Problem

Link State Routing
Link state routing is based on the assumption that, although the global knowledge about the
topology is not clear, each node has partial knowledge: it knows the state (type, condition, and
cost) of its links. In other words, the whole topology can be compiled from the partial knowledge
of each node

Building Routing Tables

1. Creation of the states of the links by each node, called the link state packet (LSP).
2. Dissemination of LSPs to every other router, called flooding, in an efficient and reliable way.
3. Formation of a shortest path tree for each node.
4. Calculation of a routing table based on the shortest path tree

I. Creation of Link State Packet (LSP) A link state packet can carry a large amount of
information. For the moment, we assume that it carries a minimum amount of data: the
node identity, the list of links, a sequence number, and age. The first two, node identity and
the list of links, are needed to make the topology. The third, sequence number, facilitates
flooding and distinguishes new LSPs from old ones. The fourth, age, prevents old LSPs from
remaining in the domain for a long time.
LSPs are generated on two occasions:
1. When there is a change in the topology of the domain
2. on a periodic basis: The period in this case is much longer compared to distance vector.
The timer set for periodic dissemination is normally in the range of 60 min or 2 h based on
the implementation. A longer period ensures that flooding does not create too much traffic
on the network.
II. Flooding of LSPs: After a node has prepared an LSP, it must be disseminated to all other
nodes, not only to its neighbors. The process is called flooding and based on the following
1. The creating node sends a copy of the LSP out of each interface
 2. A node that receives an LSP compares it with the copy it may already have. If the
newly arrived LSP is older than the one it has (found by checking the sequence number),
it discards the LSP. If it is newer, the node does the following:
a. It discards the old LSP and keeps the new one.
b. It sends a copy of it out of each interface except the one from which the packet
arrived. This guarantees that flooding stops somewhere in the domain (where a node
has only one interface).
III. Formation of Shortest Path Tree: Dijkstra Algorithm
A shortest path tree is a tree in which the path between the root and every other node is the
shortest.
The Dijkstra algorithm creates a shortest path tree from a graph. The algorithm divides the
nodes into two sets: tentative and permanent. It finds the neighbors of a current node, makes
them tentative, examines them, and if they pass the criteria, makes them permanent.
IV. Calculation of a routing table
routing table for node A

Path Vector Routing

Distance vector and link state routing are both intra domain routing protocols. They can be
used inside an autonomous system, but not between autonomous systems. These two
protocols are not suitable for inter domain routing mostly because of scalability. Both of these
routing protocols become intractable when the domain of operation becomes large. Distance
vector routing is subject to instability in the domain of operation. Link state routing needs a
 huge amount of resources to calculate routing tables. It also creates heavy traffic because of
flooding. There is a need for a third routing protocol which we call path vector routing.

Path vector routing proved to be useful for inter domain routing. The principle of path vector
routing is similar to that of distance vector routing. In path vector routing, we assume that
there is one node (there can be more, but one is enough for our conceptual discussion) in
each AS that acts on behalf of the entire AS. Let us call it the speaker node. The speaker node
in an AS creates a routing table and advertises it to speaker nodes in the neighboring ASs. The
idea is the same as for distance vector routing except that only speaker nodes in each AS can
communicate with each other. However, what is advertised is different. A speaker node
advertises the path, not the metric of the nodes, in its autonomous system or other
autonomous systems

Initialization
Initial routing tables in path vector routing

Sharing
Just as in distance vector routing, in path vector routing, a speaker in an autonomous system
shares its table with immediate neighbors. In Figure, node A1 shares its table with nodes B1
and C1. Node C1 shares its table with nodes D1, B1, and A1. Node B1 shares its table with C1
and A1. Node D1 shares its table with C1.

Updating When a speaker node receives a two-column table from a neighbor, it updates its
own table by adding the nodes that are not in its routing table and adding its own autonomous
system and the autonomous system that sent the table. After a while each speaker has a table
and knows how to reach each node in other Ass
a) Loop prevention. The instability of distance vector routing and the creation of loops can be
avoided in path vector routing. When a router receives a message, it checks to see if its AS
is in the path list to the destination. If it is, looping is involved and the message is ignored.
b) Policy routing. Policy routing can be easily implemented through path vector routing.
When a router receives a message, it can check the path. If one of the AS listed in the path
is against its policy, it can ignore that path and that destination. It does not update its
routing table with this path, and it does not send this message to its neighbors.
c) Optimum path. What is the optimum path in path vector routing? We are looking for a
path to a destination that is the best for the organization that runs the AS. One system may
use RIP, which defines hop count as the metric; another may use OSPF with minimum delay
defined as the metric. In our previous figure, each AS may have more than one path to a
destination. For example, a path from AS4 to ASI can be AS4-AS3-AS2-AS1, or it can be AS4-
AS3-ASI. For the tables, we chose the one that had the smaller number of ASs, but this is
not always the case. Other criteria, such as security, safety, and reliability, can also be
applied
Hierarchical Routing
As networks grow in size, the router routing tables grow proportionally. Not only is router
memory consumed by ever-increasing tables, but more CPU time is needed to scan them and
more bandwidth is needed to send status reports about them.
At a certain point, the network may grow to the point where it is no longer feasible for every
router to have an entry for every other router, so the routing will have to be done
hierarchically, as it is in the telephone network.
When hierarchical routing is used, the routers are divided into what we will call regions. Each
router knows all the details about how to route packets to destinations within its own region
but knows nothing about the internal structure of other regions.
For huge networks, a two-level hierarchy may be insufficient; it may be necessary to group the
regions into clusters, the clusters into zones, the zones into groups, and so on, until we run out
of names for aggregations

When a single network becomes very large, an interesting question is ‘‘how many levels
should the hierarchy have?’’
For example, consider a network with 720 routers. If there is no hierarchy, each router needs
720 routing table entries.
If the network is partitioned into 24 regions of 30 routers each, each router needs 30 local
entries plus 23 remote entries for a total of 53 entries.
If a three-level hierarchy is chosen, with 8 clusters each containing 9 regions of 10 routers,
each router needs 10 entries for local routers, 8 entries for routing to other regions within its
own cluster, and 7 entries for distant clusters, for a total of 25 entries
Kamoun and Kleinrock (1979) discovered that the optimal number of levels for an N router
network is ln N, requiring a total of e ln N entries per router

CONGESTION CONTROL ALGORITHMS

Too many packets present in (a part of) the network causes packet delay and loss that
degrades performance. This situation is called congestion.
The network and transport layers share the responsibility for handling congestion. Since
congestion occurs within the network, it is the network layer that directly experiences it and
must ultimately determine what to do with the excess packets.
However, the most effective way to control congestion is to reduce the load that the transport
layer is placing on the network. This requires the network and transport layers to work
together. In this chapter we will look at the network aspects of congestion.

When too much traffic is offered, congestion sets in and performance degrades sharply

Above Figure depicts the onset of congestion. When the number of packets hosts send into
the network is well within its carrying capacity, the number delivered is proportional to the
number sent. If twice as many are sent, twice as many are delivered. However, as the offered
load approaches the carrying capacity, bursts of traffic occasionally fill up the buffers inside
routers and some packets are lost. These lost packets consume some of the capacity, so the
number of delivered packets falls below the ideal curve. The network is now congested. Unless
the network is well designed, it may experience a congestion collapse
difference between congestion control and flow control.
Congestion control has to do with making sure the network is able to carry the offered traffic.
It is a global issue, involving the behavior of all the hosts and routers.
Flow control, in contrast, relates to the traffic between a particular sender and a particular
receiver. Its job is to make sure that a fast sender cannot continually transmit data faster than
the receiver is able to absorb it.
To see the difference between these two concepts, consider a network made up of 100-Gbps
fiber optic links on which a supercomputer is trying to force feed a large file to a personal
computer that is capable of handling only 1 Gbps. Although there is no congestion (the
network itself is not in trouble), flow control is needed to force the supercomputer to stop
frequently to give the personal computer a chance to breathe.
At the other extreme, consider a network with 1-Mbps lines and 1000 large computers, half of
which are trying to transfer files at 100 kbps to the other half. Here, the problem is not that of
fast senders overpowering slow receivers, but that the total offered traffic exceeds what the
network can handle.

The reason congestion control and flow control are often confused is that the best way to
handle both problems is to get the host to slow down. Thus, a host can get a ‘‘slow down’’
message either because the receiver cannot handle the load or because the network cannot
handle it.

Several techniques can be employed. These include:

1. Warning bit
2. Choke packets
3. Load shedding
4. Random early discard
5. Traffic shaping
The first 3 deal with congestion detection and recovery. The last 2 deal with congestion
avoidance

Warning Bit
1. A special bit in the packet header is set by the router to warn the source when congestion
is detected.
2. The bit is copied and piggy-backed on the ACK and sent to the sender.
3. The sender monitors the number of ACK packets it receives with the warning bit set and
adjusts its transmission rate accordingly.
Choke Packets
1. A more direct way of telling the source to slow down.
2. A choke packet is a control packet generated at a congested node and transmitted to
restrict traffic flow.
3. The source, on receiving the choke packet must reduce its transmission rate by a certain
percentage.
4. An example of a choke packet is the ICMP Source Quench Packet.
Hop-by-Hop Choke Packets
1. Over long distances or at high speeds choke packets are not very effective.
2. A more efficient method is to send to choke packets hop-by-hop.
3. This requires each hop to reduce its transmission even before the choke packet arrive at
the source

Load Shedding
1. When buffers become full, routers simply discard packets.
2. Which packet is chosen to be the victim depends on the application and on the error
strategy used in the data link layer.
3. For a file transfer, for, e.g. cannot discard older packets since this will cause a gap in the
received data.
4. For real-time voice or video it is probably better to throw away old data and keep new
packets.
5. Get the application to mark packets with discard priority.

Random Early Discard (RED)

1. This is a proactive approach in which the router discards one or more packets before the
buffer becomes completely full.
2. Each time a packet arrives, the RED algorithm computes the average queue length, avg.
3. If avg is lower than some lower threshold, congestion is assumed to be minimal or non-
existent and the packet is queued.
4. If avg is greater than some upper threshold, congestion is assumed to be serious and the
packet is discarded.
5. If avg is between the two thresholds, this might indicate the onset of congestion. The
probability of congestion is then calculated.
Traffic Shaping
1. Another method of congestion control is to “shape” the traffic before it enters the
network.
2. Traffic shaping controls the rate at which packets are sent (not just how many). Used in
ATM and Integrated Services networks.
3. At connection set-up time, the sender and carrier negotiate a traffic pattern (shape).

Two traffic shaping algorithms are:

Leaky Bucket
Token Bucket

The Leaky Bucket Algorithm used to control rate in a network. It is implemented as a single-
server queue with constant service time. If the bucket (buffer) overflows then packets are
discarded.

(a) A leaky bucket with water. (b) a leaky bucket with packets.
1. The leaky bucket enforces a constant output rate (average rate) regardless of the
burstiness of the input. Does nothing when input is idle.
2. The host injects one packet per clock tick onto the network. This results in a uniform flow
of packets, smoothing out bursts and reducing congestion.
3. When packets are the same size (as in ATM cells), the one packet per tick is okay. For
variable length packets though, it is better to allow a fixed number of bytes per tick. E.g.
1024 bytes per tick will allow one 1024-byte packet or two 512-byte packets or four 256-
byte packets on 1 tick
Token Bucket Algorithm

1. In contrast to the LB, the Token Bucket Algorithm, allows the output rate to vary,
depending on the size of the burst.
2. In the TB algorithm, the bucket holds tokens. To transmit a packet, the host must capture
and destroy one token.
3. Tokens are generated by a clock at the rate of one token every t sec.
4. Idle hosts can capture and save up tokens (up to the max. size of the bucket) in order to
send larger bursts later.

(a) Before. (b) After.

Leaky Bucket vs. Token Bucket

1. LB discards packets; TB does not. TB discards tokens.
2. With TB, a packet can only be transmitted if there are enough tokens to cover its length in
bytes.
3. LB sends packets at an average rate. TB allows for large bursts to be sent faster by speeding
up the output.
4. TB allows saving up tokens (permissions) to send large bursts. LB does not allow saving.