Download Document

Unit 5
The network layer
1
Topics
Network layer design issues
 Routing algorithms
 Congestion control algorithms
 Quality of service
 X.25 architecture
 Frame Relay architecture

2
Network Layer
design issues
3
Network layer design issues





Store and forward packet switching
Services provided to the transport layer
Implementation of connectionless service
Implementation of connection oriented
service
Comparison of virtual-circuit and datagram
subnets
4
Store-and-Forward Packet
Switching
The environment of the network layer protocols.
fig 5-1
5
Services provided to the
transport layer
6
Services provided to the
transport layer

The network layer services have been designed with the
following goals in mind:






1. The services should be independent of the subnet technology.
2. The transport layer should be shielded from the number, type,
and topology of the subnets present.
3. The network addresses made available to the transport layer
should use a uniform numbering plan, even across LANs and
WANs.
The network layer can be connectionless or connectionoriented
The internet has a connectionless network layer whereas
ATM network has connection oriented network layer
The network layer should provide a means to send
packets from a node A to B
7
Internal organization of the
network layer


Virtual Circuits, in analogy with the physical
circuits set up by the telephone system
Datagrams, in analogy with telegrams
8
Implementation of Connectionless
Service
Routing within a datagram subnet.
9
Implementation of ConnectionOriented Service
Routing within a virtual-circuit subnet.
Connection identifier
10
Comparison of Virtual-Circuit and
Datagram Subnets
5-4
11
Routing Algorithms
12
Routing algorithms










The optimality principle
Shortest path routing
Flooding
Distance vector routing
Link state routing
Hierarchical routing
Broadcast routing
Multicast routing
Routing for mobile hosts
Routing in Ad hoc networks
13
Routing algorithms








Routing algorithm: determine the route and
maintain the routing table
desired properties for a routing algorithm:
1. correctness
2. simplicity
3. robustness with respect to failures and changing
conditions
4. stability of the routing decisions
5. fairness of the resource allocation
6. optimality of the packet travel times
14
Routing algorithms
Fairness and optimality are often contradictory goals.
15
Routing algorithms





What is it that we seek to optimize?
Minimizing mean packet delay is an obvious
candidate, but so is maximizing total network
throughput.
Furthermore, these two goals are also in conflict,
since operating any queuing system near
capacity implied a long queuing delay.
As a compromise, many networks attempt to
minimize the number of hops a packet must
make,
because reducing the number of hops tends to
improve the delay and also reduce the amount of
bandwidth consumed, which tends to improve the
throughput as well.
16
Routing algorithms

Non-adaptive (static) algorithms:



Routing decision is not based on the estimation of
the current traffic and topology
However the choice of the network is done in
advance, offline and is downloaded to routers
Adaptive (dynamic) algorithm:

Routing decisions can be changed if there any
changes in topology or traffic
17
The optimality principal



The optimality principle states that if router J
is on the optimal path from router I to router K,
then the routes from I to J and from J to K are
also optimal.
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.
18
The Optimality Principle
(a) A subnet. (b) A sink tree for router B.
19
The Optimality Principle



A sink tree does not contain any loops, so each
packet will be delivered within a finite and bounded
number of hops.
Links and routers can go down and come back up
during operation, so different routers may have
different ideas about the current topology.
One of the issue is whether each router has to
individually acquired the information on which
to base its sink tree computation, or whether this
information is collected by some other means.
20
Shortest path routing


1.
2.
Dijikstra’s algorithm:
Let the node we are starting be called an initial node.
Let a distance of a node Y be the distance from the
initial node to it. Dijkstra's algorithm will assign some
initial distance values and will try to improve them stepby-step.
Assign to every node a distance value. Set it to zero for
our initial node and to infinity for all other nodes.
Mark all nodes as unvisited. Set initial node as current.
21
Shortest path routing
3.
4.
5.
For current node, consider all its unvisited neighbours
and calculate their distance (from the initial node).
For example, if current node (A) has distance of 6,
and an edge connecting it with another node (B) is 2,
the distance to B through A will be 6+2=8.
If this distance is less than the previously recorded
distance (infinity in the beginning, zero for the initial
node), overwrite the distance.
When we are done considering all neighbours of the
current node, mark it as visited. A visited node will not
be checked ever again; its distance recorded now is
final and minimal.
Set the unvisited node with the smallest distance (from
the initial node) as the next "current node" and
continue from step 3.
22
Shortest Path Routing
The first 5 steps used in computing the shortest path from A to D.
The arrows indicate the working node.
23
Flooding
P
flooding
P
P
P
Transmit a copy of each packet
it receives on every one of its
transmission links
advantages: robust, simple, broadcasting, discovery
disadvantages: use too much resource
How to limit the flooding: 1. hop count
2. time stamp
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.
24
Applications of flooding



Military applications
Update many databases simultaneously
In wireless networks
25
Distance Vector Routing
Distance vector routing algorithms operate by having each
router maintain a table (i.e., a vector) giving the best known
distance to each destination and which line to use to get there.
These tables are updated by exchanging information with the
neighbors.
E.g.: Routing table for Router A
Destination
B
C
D
cost(delay, distance, …)
10
12
...
via
B
B
26
Distance Vector Routing





It was the original ARPANET routing algorithm and
was also used in the Internet under the name RIP
(Routing Information Protocol)
AppleTalk and Cisco routers use improved distance
vector protocols.
Also known as Bellman-Ford algorithm and Ford
Fulkerson algorithm
Once every T msec each router sends to each
neighbor a list of its estimate delays to each
destination.
It also receives a similar list from each neighbor.
27
Distance Vector Routing



Each router maintains a routing table.
It contains one entry for each router in the subnet.
This entry has two parts:



The first part shows the preferred outgoing line to reach to
the destination
Second path gives an estimate of the time or distance to
the destination
The metric used can be one of the following:



Number of hops
Time delay
Number of packets in a queue etc,
28
Distance
Vector Routing
A subnet.
(b)
Input from A, I, H, K,
and the new routing
table for J.
(a)
29
Distance Vector Routing
The count-to-infinity problem
A is down
Then A comes up. The good news spreads quickly.
30
Distance Vector Routing
The count-to-infinity problem
A is up
Then A
comes down.
The bad
news travels
slowly.
31
Distance Vector Routing




Bad news travels slowly: no router ever has a value
more than one higher than the minimum of all its
neighbors.
Gradually, all the routers work their way up to infinity,
but the number of exchanges required depends on
the numerical value used for infinity.
For this reason, it is wise to set infinity to the
longest path plus 1 (if using hop count as metric).
If the metric is time delay, there is no well-defined
upper bound, so a high value is needed to prevent
a path with a long delay from being treated as down.
32
Link State Routing
• Each router must do the following:








Discover its neighbors, learn their network address.
Measure the delay or cost to each of its neighbors.
Construct a packet telling all it has just learned.
Send this packet to all other routers.
Compute the shortest path to every other router.
Distance vector routing differs significantly from the link
state routing.
With link state algorithms, routers share only the identity
of their neighbors, but they flood this information
through the entire network.
Distance vector algorithms adopt an opposite approach.
Routers periodically share knowledge of the entire
network, but only with their neighbors.
33
Link State Routing






Learning about neighbors
When a router is booted, its first task is to learn who
its neighbor are.
It accomplishes this goal by sending a special
HELLO packet on each point-to-point line.
The router on the other end is expected to send
back a reply telling who it is.
When two or more routers are connected by a LAN,
the situation is slighted more complicated.
One way to model the LAN is to consider it as a
node itself.
34
Link State Routing

Learning about neighbors
35
Link State Routing

Measuring line cost

Cost is measured by sending a special ECHO packet
that the other side is required to send back again.
By measuring the round trip time and dividing it by two
can give an estimate of delay
An interesting issue is whether or not to take the load
into account when measuring the delay.
To factor the load in, the round-trip timer must be started
when the ECHO packet is queued.
To ignore the load, the timer should be started when the
ECHO packet reaches the front of the queue.




36
Link State Routing
Measuring Line Cost
Including queuing cost: can use
the best line, but may lead to
routing table oscillating.
Same bandwidth
on the two links
37
Link State Routing
Building Link State Packets
Building the link state packets is easy.
The hard part is determining when to build them.
1. Periodically
2. When some significant event occurs, such as a line or neighbor
38
going down or coming back up.
Link State Routing





Distributing the link state packets
The trickiest part of the algorithm is distributing the
link state packets reliably.
As the packets are distributed and installed, the
routers getting the first ones will change their routes.
The different routers may be using different
versions of the topology, which can lead to
inconsistencies, loops, unreachable machines, and
other problems.
The fundamental idea is to use flooding to
distribute the link state packets.
39
Link State Routing





Distributing the link state packets
To keep the flood in check, each packet contains a
sequence number that is incremented for each new
packet sent. Routers keep track of all the (source
router, sequence) pairs they see.
When a new link state packet comes in, it is
checked against the list of packets already seen.
1. If new: forward on all lines except the one it
arrived on
2. If duplicate or old packet: discard
40
Link State Routing




Distributing the link state packets
Problems:
If the sequence numbers wrap around,
confusion will reign. The solution here is to use a
32-bit sequence number. With one link state
packet per second, it would take 137 years to
wrap around.
If a router crashes, it will lose track of its
sequence number.
41
Link State Routing

Distributing the link state packets

The solution to all these problems is to include the
age of each packet after the sequence number and
decrement it once per second. When the age hits
zero, the information from that router is discarded.

The age field is also decremented by each router
during the initial flooding process, to make sure no
packet can get lost and live for an indefinite period
of time.
42
Link State Routing
Distributing the Link State Packets
To guard against errors on the router-router lines, all link state
packets are acknowledged.
Packet buffer for router B
43
Link State Routing

Computing the new routes

Once a router has accumulated a full set of link
state packets, it can construct the entire subnet
graph because every link is represented.
Now Dijkstra’s algorithm can be run locally to
construct the shortest path to all possible
destinations.
The OSPF (Open Shortest Path First) protocol uses
link state routing algorithm.


44
Hierarchical routing

Cx from U.S. send a packet to csie.ndhu.edu.tw






1. first send to domain tw (Taiwan)
2. then to subdomain MOE (edu)
3. then to subsubdomain ndhu
4. then to host csie
Advantage: simple, efficient, and saving routing
table space
In a 1000 users network, each routing table needs
999 entries. If it is divided into 10 domains, then
each table only needs 99+9=108 entries.
45
Hierarchical
routing
46
Hierarchical routing




Unfortunately, the gain in routing table space are not
free. There is a penalty to be paid, and this penalty
is in the form of increased path length.
For example, the best route from 1A to 5C is via
region 2, but with hierarchical routing all traffic to
region 5 goes via region 3, because that is better for
most destinations in region 5.
When a single network becomes very large, an
interesting question is: How many levels should the
hierarchy have?
Answer: the optimal number of levels for an N router
subnet is logN,
47
Broadcast routing

One broadcasting method that requires no special
features from the subnet is for the source to simply
send a distinct packet to each destination.

Waste bandwidth and require the source to have a
complete list of all destinations.

Flooding is another obvious candidate. But it
generates too many packets and consumes too
much bandwidth.
48
Broadcast routing



Multidestination routing
Each packet contains either a list of destinations or
a bit map indicating the desired destinations. When
a packet arrives at a router, the router checks all the
destinations to determine the set of output lines that
will be needed.
The router generates a new copy of the packet for
each output line to be used and includes in each
packet only those destinations that are to use the
line. In effect, the destination set is partitioned
among the output lines.
49
Broadcast routing

A fourth broadcast algorithm makes explicit use of the
sink tree for the router initiating the broadcast, or any
other convenient spanning tree for that matter.

This method makes excellent use of bandwidth,
generating the absolute minimum number of packets
necessary to do the job. The only problem is that each
router must have knowledge of some spanning tree for it
to be applicable.
50
Broadcast routing


Reverse path forwarding
When a broadcast packet arrives at a router, the router
checks to see if the packet arrived on the line that is
normally used for sending packets to the source of the
broadcast. If so, forward it.
Reverse path forwarding. (a) A subnet. (b) a Sink tree. (c) The
tree built by reverse path forwarding.
51
Multicast routing





To do multicasting, group management is required.
Some way is needed to create and destroy groups,
and for processes to join and leaves groups.
It is important that routers know which of their
hosts belong to which groups.
Either hosts must inform their routers about changes
in group membership, or routers must query their
hosts periodically.
Either way, routers learn about which of their hosts
are in which groups.
Routers tell their neighbors, so the information
propagates through the subnet.
52
Multicast routing

To do multicast routing, each router computes a spanning tree
covering all other routers in the subnet.
(a) A network. (b) A spanning tree for the leftmost router.
53
(c) A multicast tree for group 1. (d) A multicast tree for group 2.
Multicast routing

Various ways of pruning the spanning tree are
possible. The simplest one can be used if link state
routing is used, and each router is aware of the
complete subnet topology, including which hosts
belong to which groups.

Then the spanning tree can be pruned by starting
at the end of each path and working toward the
root, removing all routers that do not belong to the
group in question.
54
Multicast routing

With distance vector routing, whenever a router with
no hosts interested in a particular group and no
connections to other routers receives a multicast
message for that group, it responses with a PRUNE
message, telling the sender not to send it any more
multicasts for that group.

source-specific multicast trees: scales poorly to
large networks n groups, m members: a total of nm
trees

core-based tree approach: each group has only
one multicast tree n groups: n trees
55
Routing for mobile hosts
A WAN to which LANs, MANs, and wireless cells are attached.
56
Routing for mobile hosts

When a new user enters an area, either by
connecting to it, or just wandering into the cell, his
computer must register itself with the foreign agent
there. The registration procedure typically works like
this:

1. Periodically, each foreign agent broadcasts a
packet announcing its existence and address. A
newly arrived mobile host may wait for one of these
messages, but if none arrives quickly enough, the
mobile host can broadcast a packet saying: “Are
there any foreign agents around?”
57
Routing for mobile hosts

2. The mobile host registers with the foreign agent,
giving its home address, current data link layer
address, and some security information.

3. The foreign agent contacts the mobile host’s
home agent and says: “One of your hosts is over
here.” The message from the foreign agent to the
home agent contains the foreign agent’s network
address. It also includes the security information, to
convince the home agent that the mobile host is
really there.
58
Routing for mobile hosts

4. The home agent examines the security information,
which contains a time stamp, to prove that it was
generated within the past few seconds. If it is happy, it
tells the foreign agent to proceed.

5. When the foreign agent gets the acknowledgement
from the home agent, it makes an entry in its tables and
informs the mobile host that it is now registered.

Ideally, when a user leaves an area, that, too, should be
announced to allow deregistration, but many users
abruptly turn off their computers when done.
59
Routing for mobile hosts
Packet routing for mobile hosts
60
Routing in Ad Hoc Networks
Possibilities when the routers are mobile:
 Military vehicles on battlefield.


A fleet of ships at sea.


All moving all the time
Emergency works at earthquake .


No infrastructure.
The infrastructure destroyed.
A gathering of people with notebook computers.

In an area lacking 802.11.
61
Route Discovery
(a) Range of A's broadcast.
 (b) After B and D have received A's broadcast.
 (c) After C, F, and G have received A's broadcast.
 (d) After E, H, and I have received A's broadcast.
Shaded nodes are new recipients. Arrows show possible reverse
62
routes.

Route Discovery
Format of a ROUTE REQUEST packet.
Format of a ROUTE REPLY packet.
63
Route Maintenance
(a) D's routing table before G goes
down.
(b) The graph after G has gone down.
(a) D's routing table before G goes down.
(b) The graph after G has gone down.
64
Congestion control
algorithms
65
Congestion control algorithms






General principles of congestion control
Congestion prevention policies
Congestion control in virtual-circuit subnets
Congestion control in datagram subnets
Load shedding
Jitter control
66
Congestion
When too much traffic is offered, congestion sets in and
performance degrades sharply.
67
Common causes of
congestion




As packets arrive at a node, they are stored in
the input buffer. If packets arrive too fast an
incoming packet may find that there is no
available buffer space
Even very large buffer cannot prevent
congestion due to delay, timeout and
retransmissions
Slow processors may lead to congestion
Low bandwidth lines may also lead to
congestion
68
Congestion




If packet arrival rate exceeds the packet
transmission rate, the queue size grows without
bound
Delay in delivery of packets leads to retransmissions
When the line for which packets are queuing
becomes more than 80% utilized the queue length
grows alarmingly
When too many packets arrive at a part of packet
switched network, the performance degrades. This
is known as congestion
69
General principles of
congestion control


Congestion control refers to technique which
can remove congestion from happening or
which can remove congestion after it has
taken place
Two categories:


Open loop: based on prevention of congestion
Closed loop: used for removing the congestion
70
General principles of
congestion control

Open loop control:



Try to solve the problems by excellent design to
prevent congestion from happening
Open loop congestion control is exercised by
using tools such as deciding when to accept the
new packets, when to discard the packets, which
packet are to be discarded and making
scheduling decisions at various points
Includes retransmission policy, window policy,
acknowledgement policy, discarding policy,
admission policy
71
General principles of
congestion control
Closed loop congestion control:
Uses feedback mechanism
It contains following steps:
Monitor the system .



1.

2.
3.
detect when and where congestion occurs.
Pass information to where action can be taken.
Adjust system operation to correct the problem.
72
Congestion prevention
policies
73
Congestion Control in VirtualCircuit Subnets






One of congestion control technique is admission
control
This technique is used to keep the congestion which
has already begun to a manageable level
Its principle is: once congestion has been detected
do not set up any more virtual circuits until
congestion has been cleared
The advantage is that it is simple and easy to carry
out control
Alternative approach: allow virtual circuits to set up
even when a congestion has taken place
However carefully route all the new virtual circuits
around the problem area
74
Congestion Control in Virtual-Circuit
Subnets
(a) A congested subnet. (b) A redrawn subnet, eliminates
congestion and a virtual circuit from A to B.
75
Congestion Control in VirtualCircuit Subnets






Another strategy is to negotiate an agreement
between the host and the subnet when a virtual
circuit is set up
The agreement specifies volume and shape of
traffic, QoS required, etc.
To keep part of the agreement, the subnet will
typically reserve resources along the path when the
circuit is set up
These resources can include table and buffer space
in routers and bandwidth on lines
In this way congestion is unlikely to occur
This may lead to underutilization of resources
76
Congestion Control in datagram
Subnets





Can be used in virtual circuits as well
In this technique each router associates a real
variable with each of its output lines. This real
variable say “u” has a value between 0 and 1 and
indicates the percentage of utilization of that line.
If the value of “u” goes above threshold then that
output line will enter into a “warning” state.
The router will check each arriving packet to see if
its output line is in “warning state”
If it is in warning state, following action is taken.


Using warning bit
Using choke packets
77
Congestion Control in
datagram Subnets

The warning bit:

Warning state denoted by setting a special bit in the
packet’s header.
When a packet is arrived at its destination, the transport
entity copied the bit into the next acknowledgement sent
back to the source
The source then cut back on traffic
Source monitors the fraction of acknowledgement with
the bit set and adjusted its transmission accordingly
As long as the warning bits continued to flow in, the
source continued to decrease its transmission rate
When they slowed, source increases its transmission
rate





78
Congestion Control in
datagram Subnets






Each node monitors the utilization of the output lines
If the utilization goes beyond some threshold level, an
output line enters a “warning state”
If the output line of the newly arriving packet is in the
warning state, a control packet called choke packet is
sent from the congested node to the source station
The original packet is tagged so that it does not generate
any more choke packets
After receiving a choke packet the source station
reduces the traffic by certain percentage.
Initially reduces traffic by 50% then by 25% and so
on…..
79
Hop-by-Hop
Choke Packets
(a) A choke packet that affects
only the source.
(b) A choke packet that affects
each hop it passes through.
80
Load shedding






When other methods fail, the nodes can resort to a
heavy artillery: Load shedding
When the nodes are not able to handle the packets, the
packets are discarded
A node drowning in packets, one approach is to discard
packets at random
In many situations, some packets are more important
than others
To implement this intelligent discard policy, the
applications must mark the packets with priority classes
Policy for file transfer is called wine (old is better than
new) and that for multimedia is called milk (new is better
than old)
81
Jitter






Jitter is defined as the variation in delay for packets
belonging to the same flow.
The real time audio and video cannot tolerate jitter on the
other hand
The jitter does not matter if the packets are carrying
information contained in a file
For the audio and video if the video transmission of the
packets take 20 msec to 30 msec to reach to the
destination, it doesn’t matter , provided the delay remains
constant
The quality of delay will be hampered if the delays
associated with the different packets have different values
An agreement that 99% packets can be delivered with a
delay ranging from 24.5 msec to 25.5 msec is acceptable
82
Jitter Control
(a) High jitter.
(b) Low jitter.
83
Jitter Control




When a packet arrives at a router, the router will
check to see whether the packet is ahead or behind
and by what time
This information is stored in the packet and is
updated at every hop
If the packet is ahead of schedule (early) the router
will hold it for a slightly longer period of time, and if
the packet is behind the schedule (late) then the
router will try to send it as quickly as possible
This will help in avoiding the average delay per
packet constant and will avoid time jitter
84
Quality of service
85
Quality of service





Requirements
Techniques for achieving good quality of
service
Integrated services
Differentiated services
Label Switching and MPLS
86
Requirements




A stream of packets is called flow.
The needs of each flow may follow different routes.
The needs of flow may be characterised by 4 primary parameters:
reliability, delay, jitter and bandwidth
Together these determine the QoS the flow requires
87
Techniques to achieve good
quality of service









Overprovisioning
Buffering
Traffic shaping
The Leacky bucket algorithm
The token bucket algorithm
Resource reservation
Admission control
Proportional routing
Packets scheduling
88
Overprovisioning


Provide much router capacity, buffer space
and bandwidth so that the packets just fly
through the network easily.
It is expensive
89
Buffering
Smoothing the output stream by buffering packets.
•
•
Flows can be buffered on the receiving side before
being delivered
Buffering does not affect reliability or bandwidth,
increases delay, buts smoothes out jitter
90
Traffic shaping






One of the reason behind congestion is the bursty
nature of the traffic. If the traffic has a uniform data
rate then congestion could be less common
Traffic shaping is open loop control.
It manages the congestion by forcing the packet
transmission rate to be more predictable
Thus traffic shaping will regulate the average rate or
burstiness of data transmission
Check if a packet stream obeys its descriptor, and if
it violates its descriptor, give penalty
For this the network may want to monitor the traffic
flow during the connection period.
91
Traffic shaping








The process of monitoring and enforcing the traffic
flow is called traffic policing
Penalty will be:
Drop the packets that violate the descriptor
Give low priority to them
Traffic shaping is a mechanism to control the
amount and rate of the traffic sent to the network
The two types of traffic shaping techniques are:
Leaky bucket
Token bucket
92
The Leaky Bucket Algorithm
(a) A leaky bucket with water. (b) a leaky bucket with packets.
93
The Leaky Bucket Algorithm






Used to control congestion in network traffic
Leaky bucket is a bucket with a hole at bottom.
Flow of the water from the bucket is at a constant
rate which is independent of water entering the
bucket
If the bucket is full, any additional water entering in
the bucket is thrown out
Same technique is applied to congestion control
Every host in the network is having a buffer with
finite queue length
94
The Leaky Bucket Algorithm





Packets which are put into the buffer when buffer is
full are thrown away.
The buffer may drain onto the subnet either by some
number of packets per unit time, or by some total
number of bytes per unit time.
A FIFO queue is used for holding the packets
If the arriving packets are fixed in size, then the front
of queue removes a fixed number of packets from
the queue at each tick of the clock
If the arriving packets are of variable size, then the
fixed output rate will not be based on number of
departing packets, instead on the number of
departing bits
95
The Token Bucket Algorithm
5-34
(a) Before.
(b) After.
96
The Token Bucket Algorithm







Similar to leaky bucket but allows varying traffic
A token generator constantly produces tokens at the
rate of one token every ΔT sec
Every time when a packet is transmitted a token is
destroyed and if the number of tokens are over then
no packets can be transmitted
The token bucket can be easily implemented with a
counter.
The token is initialized to zero.
Each time the token is added, the counter is
incremented by 1 and each time a unit of data is
dispatched the counter is decremented by 1
If the counter contains zero no data can be
transmitted
97
The Leaky
Bucket
Algorithm
(a) Input to a leaky bucket.
(b) Output from a leaky
bucket. Output from a token
bucket with capacities of (c)
250 KB, (d) 500 KB, (e)
750 KB, (f) Output from a
500KB token bucket feeding
a 10-MB/sec leaky bucket.
98
Resource reservation




The data flow is dependent on the following
resources:
i) Buffer ii) bandwidth and iii) CPU time
The QoS can be improved by reserving these
resources
The QoS model called integrated services
depends heavily on the principle of resource
reservation for improvement in QoS
99
Admission control




Admission control technique is
used by the router or switch
They use this mechanism to accept
or reject the flow based on
predefined parameters called flow
specifications
Before accepting a flow for
processing, a router checks to the
flow specifications to judge if it can
handle this new data flow
It does this by checking its capacity
in terms of bandwidth, buffer size,
CPU speed, etc and its
commitments to other flow
An example of flow specification
100
Proportional routing




Most routers use best path to send all traffic
to the destination
A different approach is to split the traffic for
each destination over multiple paths
Can use locally available information
Divide the traffic equally to the proportion of
the outgoing links
101
Packet Scheduling
(a) A router with five packets queued for line O.
(b) Finishing times for the five packets.
102
Packet scheduling





Each router must implement some queuing
discipline which governs how the packets are
buffered when they are waiting to get transmitted
If a router is handling multiple flows, there is a
danger that one flow will take too much of its
capacity and starve all other flows
Two algorithms are used for this:
Fair queuing: queues are scanned in a round robin
manner. Packets are sorted in the order of their
finishing time and sent in that order
Weighted fair queuing: Instead of giving same
priority to all the hosts, higher priority can be given
to some hosts, such as server leading to modified
algorithm called weighted fair queuing
103
Integrated services



Used for multimedia streams
Also known as flow-based algorithms
Used for multicast applications
104
Resource reSerVation protocol





Multicast flows from multiple source to
multiple destinations
Uses multicast routing using spanning trees
Each group is assigned a group address
which is used to send packets to that group
The multicast routing algorithm builds a
spanning tree covering all group members
To eliminate congestion any of the receivers
can send reservation message up the tree to
the sender
105
RSVP-The ReSerVation Protocol
(a) A network, (b) The multicast spanning tree for host 1.
(c) The multicast spanning tree for host 2.
106
RSVP-The ReSerVation Protocol
(a) Host 3 requests a channel to host 1. (b) Host 3 then requests a
107
second channel, to host 2. (c) Host 5 requests a channel to host 1.
Differentiated services (DS)






DS can be offered by a set of routers forming an
administrative domain
The administration defines a set of service classes with
corresponding forwarding rules:
If a customer signs up for DS, customer packets entering
the domain may carry a type of Service field in them with
a better service provided to some classes than to others
Traffic within a class may be required to conform to some
specific shape, such as leaky bucket with some specified
drain rate
Also known as class based service
Can be implemented using
 Expedited forwarding
 Assured forwarding
108
Differentiated services (DS)




Example of Internet telephony:
With a flow-based scheme, each telephone
call gets its own resources and guarantees
With a class based scheme, all telephone
calls together get the resources reserved for
the class telephony.
These resources cannot be taken away by
packets from file transfer or other classes, but
no telephone call gets any private resources
reserved for it alone
109
Expedited Forwarding
Expedited packets experience a traffic-free
network.
110
Assured Forwarding







Manages service classes
It specifies that there shall be four priority classes,
each class having its own resource
It defines three discard probabilities for packets that
are undergoing congestion: low, medium and high
Packets processed under assured forwarding is as
follows:
Step 1: classify the packets into four priority classes
Step 2: Mark the packets according to their classes
Step 3: Pass the packets through a shaper/ dropper
that may delay some of them into acceptable forms
111
Assured Forwarding
A possible implementation of the data flow for
assured forwarding.
112
Label Switching and MPLS
Transmitting a TCP segment using IP, MPLS,
and PPP.
113
Label Switching and MPLS



A technique of adding a label and doing the
routing based on that label in front of each
packet instead of destination address is
known as label switching
Using this technique routing can be done
quickly and any necessary resources can be
reserved along the path
A label called MPLS (MultiProtocol label
switching) is created
114
The X.25 network




To use X.25 a computer first established a
connection to the remote computer, that is placed a
telephone call.
This connection was given a connection number to
be used in data transfer packets
Data packets were very simple, consisting of a 3byte header and up to 128 bytes of data.
The header consisted of a 12-bit connection
number, a packet sequence number, an
acknowledge number, and a few miscellaneous
bits.
115
X.25 network

Designed to provide a low cost alternative for data
communication over public networks




Pay only for bandwidth actually used
Ideal for “bursty” communication over low quality
circuits
Standard provides error detection and correction for
reliable data transfer
X.25 standard approved in 1976 by CCITT (now known
as ITU)


Can support speeds of 9.6 Kbps to 2 Mbps
Can provide multiplexing of up to 4095 virtual
circuits over on DTE-DCE link
116
X.25 Devices

Data Terminal Equipment (DTE)



Data Circuit-terminating Equipment (DCE)



Terminals, personal computers, and network
hosts
Located on premises of subscriber
Modems and packet switches
Usually located at carrier facility
Packet Switching Exchange (PSE)

Switches that make up the carrier network
117
Sample X.25 Network
PSE
X.25
WAN
PSE
Modem
DCE
Terminal
DTE
Personal Computer
DTE
Modem
DCE
PSE
PSE
Modem
DCE
Server
DTE
118
Frame relay






Virtual circuit WAN
Prior to frame relay X.25 was used
X.25 had following problems:
X.25 has low 64 kbps data rate
Intensive flow control and error control at
data link and network layer which slows down
the network and also a overhead
Originally used for private networks, not for
internet so has its own network layer
119
Frame relay features








Operates at higher speeds (1.554 Mbps and 44.376)
Operates in physical and data link layers
Acts as backbone networks
Allows bursty data
Allows frame size of 9000 bytes, which can
accommodate all local area networks frames
Cheap compared to others
Error detection at data link layer only
No flow control or error control at higher layers
120
Architecture

Frame relay network
Packet Switch
DCE
Packet Switch
DCE
Personal Computer
DTE
Terminal
DTE
Frame Relay
WAN
Packet Switch
DCE
Packet Switch
DCE
Network Host
DTE
121
Frame Relay Devices

Devices attached to a Frame Relay WAN fall into
the following two general categories:


Data terminal equipment (DTE)
 DTEs generally are considered to be terminating equipment
for a specific network and typically are located on the
premises of a customer.
 Example of DTE devices are terminals, personal computers,
routers, and bridges.
Data circuit-terminating equipment (DCE)
 DCEs are carrier-owned internetworking devices.
 The purpose of DCE equipments is to provide clocking and
switching services in a network, which are the devices that
actually transmit data through the WAN.
122
Frame Relay Virtual Circuits




Provides connection-oriented data link layer
communications.
Defined communication path exists and connections
associated with a connection identifier (ID).
Implemented by using a FR virtual circuit, which is a
logical connection created between two DTE
devices across a Frame Relay packet-switched
network (PSN).
Virtual circuits provide a bidirectional communication
path from one DTE device to another and uniquely
identified by a data-link connection identifier (DLCI).
123
Frame Relay Virtual Circuits
(cont.)




A number of virtual circuits can be multiplexed into a
single physical circuit for transmission across the
network.
This capability often can reduce the equipment and
network complexity required to connect multiple
DTE devices.
A virtual circuit can pass through any number of
intermediate DCE devices (switches) located within
the Frame Relay PSN.
Frame Relay virtual circuits fall into two categories:
switched virtual circuits (SVCs) and permanent
virtual circuits (PVCs).
124
Switched Virtual Circuits (SVCs)



Temporary connections
used in situations requiring only sporadic data transfer
between DTE devices.
A communication session across an SVC consists of the
following four operational states:




Call setup—The virtual circuit between two Frame Relay
DTE devices is established.
Data transfer—Data is transmitted between the DTE
devices over the virtual circuit.
Idle—The connection between DTE devices is still active,
but no data is transferred. If an SVC remains in an idle
state for a defined period of time, the call can be
terminated.
Call termination—The virtual circuit between DTE devices
is terminated.
125
Permanent Virtual Circuits (PVCs)






Permanently established connections used for
frequent and consistent data transfers between DTE
devices.
Does not require the call setup and termination states
PVCs always operate in one of the following two
operational states:
Data transfer—Data is transmitted between the DTE
devices over the virtual circuit.
Idle—The connection is active, but no data is
transferred. Unlike SVCs, PVCs will not be terminated
under any circumstances when in an idle state.
DTE devices can begin transferring data whenever
they are ready because the circuit is permanently
established.
126
Data-Link Connection Identifier (DLCI)



Frame Relay virtual circuits are identified by data-link
connection identifiers (DLCIs).
DLCI values typically are assigned by the Frame Relay
service provider (for example, the telephone company).
Frame Relay DLCIs have local significance, which
means that their values are unique in the LAN, but not
necessarily in the Frame Relay WAN.
127
Frame Relay switches


Each switch in a frame relay network has a
table to route frames.
The table matches an incoming port DLCI
with an outgoing port DLCI combination
128