Traffic Management

1
 Traffic Management
• Congestion Overview
– Congestion can occur anytime the amount of data that needs to be
transmitted by a particular media exceeds the bandwidth of that
media.
– Congestion anywhere in the path results in delays for user
applications
– Periodic congestion often occurs due to the busty nature of today’s
network applications and some temporary congestion is to be
expected in each network. Continual congestion or slowness is
not normal and the causes should be determined.

2
 Traffic Management
• Traffic in an IP Network
– Data: Examples are
• File transfer: FTP and TFTP protocols
• e-mail: SMTP protocol
– Overhead: Examples are
• Routing protocol updates
• Broadcast requests, such as for a Domain Name Server (DNS)
• The use of Address Resolution Protocol (ARP) to resolve logical-to-
physical addressing issues.
• Traffic Congestion is Caused by the Following:
– Bursts of user application traffic
– Multicast and broadcast traffic
– Over-utilized low bandwidth links
– Network design issues

3
 Traffic Management
• Network Congestion Can be Controlled in the Router Through the Use of:
– Adjusting timers on periodic announcements
• To lengthen the interval between the broadcast by adjusting timers
reduces the overall traffic load on the link.
– Providing static entries in routing tables.
• The use of static entries in a routing table can obviate the need to
dynamically advertise network routes across that link. This technique
is very effective for serial lines.
– Applying standard access lists
• Standard access lists usually filter traffic based upon source
addressing characteristics. It can prevent irrelevant traffic from
reaching critical links.
– Priority and queues
• Reorder application traffic flowing across a serial link in a priority
queue so that all traffic of a particular type gets through first, or in a
queue where traffic gets a certain percentage of the bandwidth.

4
 Traffic Management
• Access List
– An access list is a series of rules that control the traffic that flows into or
out of an interface of a router. An access list is a sequential collection of
permits and deny conditions that apply to IP addresses.
– The router tests addresses against the conditions in an access list one by
one.
– The first match determines whether the router accept or rejects the
address.
– Because the router stops testing conditions after the first match, the order
of the conditions is critical.
– If no condition match, the router reject the address.

5
 Traffic Management
Inbound Access List Processing
No
Incoming packet Access list?

Yes

Does source address

Next entry in list
match?
Yes
No

More
Yes entries? Apply condition

No

Deny Permit Route to interface

and forward packet

Issue ICMP Host

Unreachable
Message

Note: For inbound access lists, after receiving a packet, the router checks the source address of the packet against the
access list
If the access list permits the address, the router continues to process the packet. If the access list rejects the address, the
router discards the packet and returns an ICMP Host Unreachable message.

6
 Traffic Management

• The syntax of an entry in a standard access list (CISCO IOS)

#access-list number action source
The parameters are:
number: A number between 1 and 99, identifying the list for future reference
action: The keyword permit or deny, indicating whether to allow or block
the packet
source: The packet’s source address
The following table shows three ways to specify the source
addresses. In most cases, the address/mask pairs is used to specify
blocks of addresses.
Source Syntax Example Explanation

address mask 192.168.2.0 0.0.0.255 Describes a block of IP address. The mask is used as a wildcard;
a one(1) in the mask indicates that we “do not care” about the
corresponding bit in the address. A zero (0) in the mask means
that the corresponding bit must match exactly. This example
specifies address from 192.168.2.0 to 192.168.2.255. That is, the
wildcard mask says that we ignore the last byte of the address
when figuring out whether an address matches
host address host 192.168.2.1 The given address must be matched exactly.

any any any IP address will match.

7
 Traffic Management
• Examples of Access List Processing
– Example 1
#access-list 1 deny 10.10.1.0 0.0.0.255
#access-list 1 deny 10.10.2.0 0.0.0.255
#access-list 1 permit any
Note:
The router processes each line in order until it finds a match.
Therefore, if a packet arrives from 10.10.2.13, it matches the second
rule in the list and so is denied.
– Example 2
#access-list 1 permit any
#access-list 1 deny 10.10.1.0 0.0.0.255
#access-list 1 deny 10.10.2.0 0.0.0.255
Note:
The first line permits all traffic because all incoming packets match it.
The second and third lines are never used. Therefore, access lists
must be ordered carefully.

8
 Traffic Management

Traffic Management is concerned with delivery of QoS to

the end user and with efficient use of network resources.

Based on traffic granularity traffic management is classified

into three levels:-
• Packet level
• Flow level
• Flow aggregated level

9
Traffic management at Packet level

10
 Traffic Management
• FIFO (first in, first out)
– The first packet received is the first to be sent out
– Delay and loss experienced by packets in FIFO queuing system depends
on packet interarrival times and on packet length
– The main problem with FIFO is that it does not separate packets according
to the queue to which they belong. Application, such as telephony, could
flood the routers with its own packets, thereby causing other application’s
packets to be discarded.
• Priority queuing
– A very stringent algorithm that can cause one type of traffic to monopolize
available bandwidth, because as long as there are high-priority packets in
the queue, they’ll be processed first. Other traffic is processed only when
there’s available bandwidth left over from high-priority traffic.
• Fair queuing
– Share equally with all traffic but gives low volume traffic higher
priority. Instead of assigning priorities to each packet, this algorithm
tracks the session that a packet belongs to. There is no queue list to
configure or apply to the interface.
– Three different systems
• Ideal fluid flow fair queuing system
• Packet-by packet fair queuing system
• Packet-by-packet weighted fair queuing system
11
 Traffic Management
• Quality of Service (QoS)
– A set of metrics used to measure the quality of transmission
and service available of any given transmission system.
– A guaranteed throughput level for critical network application.
QoS parameters are used in traffic engineering to state the level
of loss (inverse of throughput), latency, and jitter that a traffic
stream will be guaranteed in a network.
• Queue:
– Broadly, any list of elements arranged in an orderly fashion and
ready for processing.
– In routing, it refers to a backlog of information packets waiting
in line to be transmitted over a router interface.

12
Figure 7.41 (a) FIFO queuing; (b) FIFO queuing with discard priority

(a)
Packet buffer
Arriving
packets
Transmission
Packet discard link
when full

(b)
Packet buffer
Arriving
packets
Transmission
link
Class 1
discard Class 2
when full discard
when threshold
exceeded

13
Figure 7.42 Head-of-line (HOL) priority queueing

Packet discard
when full

High-priority
packets Transmission link

Low-priority When
packets high-priority
queue empty
Packet discard
when full

14
 Figure 7.43 Sorting packets according to priority tag

Sorted packet buffer

Arriving
Tagging
packets
unit
Transmission
Packet discard link
when full

15
 Traffic Management
• Ideal fluid flow fair queuing system
– Transmission bandwidth is divided equally among all nonempty
queues. (For example, if the total number of flows in the system
is n and the transmission capacity is C, then each flow is
guaranteed at least C/n (bits/second)
– One approach could be to service each nonempty queue one bit
at a time in round-robin fashion

16
 Figure 7.44 Ideal fluid flow system

Approximated
Packet flow 1 bit-level
round robin
Packet flow 2 service C bits/second

Transmission link
Packet flow n

17
 Traffic Management

• Packet-by-packet fair queueing system

– One approach could be to service each nonempty queue one
packet at a time in round-robin fashion.
• This approach is not really fair. For example, if the packets
of one flow are twice the size of packets in another flow,
then in the long run the first flow will obtain twice the
bandwidth of the second flow.
– A better approach is based on finish tag concept. The goal of
this approach is to to have each packet’s completion time
approximate that of a ideal fluid flow fair queuing system
• Each time a packet arrives at a queue, the completion time
of the packet is derived from an ideal fluid flow fair queuing
system. The number is used as a finish tag for the packet.
• Each time the transmission of a packet is complete. The next
packet to be transmit is the one with smallest finish tag
among all of the queues.
18
 Figure 7.45 Ideal fluid flow system and packet-by-packet fair queuing system
(two packets of equal length)

Queue 1 Ideal fluid flow system:

@ t=0 both packets served
at rate 1/2
Queue 2 1
Both packets
@ t=0
complete service
at t=2
t
0 1 2

Packet from
Packet-by-packet queueing system:
queue 2 waiting
queue 1 served first at rate 1;
then queue 2 served at rate 1.
1
Packet from queue 2
being served
Packet from t
queue 1 being 0 1 2 19
served
 Traffic Management
• Round (in terms of ideal fluid flow fair queuing system)
– A round consists of a cycle in which all n queues are offered
service, one bit at a time
– The actual duration of a given round is the actual number of
queues nactive (t) that have information to transmit.
• When the number of active queues is large, the duration of a
round is large
• When the number of active queue is small, the duration of a
round is small
– Round is in unit of bit

20
 Traffic Management
• Meaning of the equation dR(t)/dt = C/nactive(t)
– Given a ideal fluid flow fair queuing system, and given the fact
that the system started at t = 0
– Let R(t) be the number of the rounds at time t, that is, the number
of cycles of services to all n queues. Assuming R(t) is a
continuous function, then
dR(t)/dt = C/nactive(t)
– Interpretation of the equation above
• For a given duration and number of active queues, the higher
the transmission capacity in bits/second, the more rounds can
be completed.
• For a given duration and transmission capacity, more active
queues means less round can be completed.

21
 Figure 7.46 Computing the finishing time in packet-by-
packet fair queueing and weighted fair queueing

Rounds Generalize so R(t) is continuous, not discrete

R(t) grows at rate inversely

proportional to nactive(t)

22
 Traffic Management

• Packet-by-packet weighted fair queuing system

– Because different users have different requirements, each user
flow has a weight that determines its relative share of the
bandwidth.

23
Figure 7.47 Ideal fluid flow system and packet-by-packet fair queuing system
(two packets of different lengths)

2 Ideal fluid flow system:

Queue 1
both packets served
@ t=0
at rate 1/2
Queue 2 1
Packet from queue
@ t=0
s served at rate 1

t
0 2 3

Packet from Packet-by-packet fair queueing:

queue 2 waiting queue 2 served at rate 1

1
Packet from
queue 1 being
served at rate 1 t
1 2 3 24
0
 Figure 7.48 Ideal fluid flow system and Packet-by-packet weighted fair
queuing system

Queue 1 Ideal fluid flow system:

@ t=0 packet from queue 1
served at rate 1/4;
Queue 2
1
@ t=0 Packet from queue 1
served at rate 1
Packet from queue 2 t
served at rate 3/4 0 2
1

Packet from
Packet-by-packet weighted fair queueing:
queue 1 waiting
queue 2 served first at rate 1;
then queue 1 served at rate 1.
1
Packet from queue 1
Packet from being served
queue 2 being
served t
0 1 2
25
 Traffic Management

• Calculation of finish tags

– Notation
• k: k-th packet
• i: i-th flow
• C: Capacity in bits/second
• P(i,k): length of k-th packet from i-th flow in bits
• R(t): the number of rounds at time t in bits
• F(i,k): finish tag of k-th packet for i-th flow

– Case 1 Empty queue: Suppose that k-th packet from flow i arrives at an
empty queue at time tki and suppose that the packet has length P(i,k), then
F(i,k) = R(tki )+ P(i,k)

finish tag bits completed at time tki packet length in bits

26
 Traffic Management
• Calculation of finish tags
– Case 2 Non-empty queue: Suppose that k-th packet from flow i arrives at
an non-empty queue at time tki and suppose that the packet has length
P(i,k), then
F(i,k) = F(i, k-1) + P(i,k)

finish tag finish tag of the previous packet in queue packet length in bits

– General case for packet-by-packet fair queueing system

F(i,k) = max {F(i, k-1), R(tki) } + P(i,k)

27
 Traffic Management
• Calculation of finish tags
– General case for packet-by-packet weighted fair queueing system
F(i,k) = max {F(i, k-1), R(tki) } + P(i,k)/wi

where wi is the weight of i-th flow

28
 Traffic Management
• Example ( Chapter 7 problem 51)
Problem statement: consider a packet-by-packet fair queuing system
with three logical queues and with service rate of one unit per second.
Show the sequence of transmission for this system for the following
packet arrival pattern.
Queue 1: arrival at time t = 0, length 2; arrival at t = 4, length 1;
Queue 2: arrival at time t = 1, length 3; arrival at t = 2, length 1.
Queue 3: arrival at time t = 3, length 5.

29
 Traffic Management

Queue Packets’ arrival F(queue number, packet number) Order of

time and length transmission

Queue 1 Packet 1: t = 0, F(1,1) = R(0) + 2 = 0 + 2 = 2 1

L=2
Packet 2: t = 4, F(1,2) = max{F(1,1), R(4)} + 1 = 4
L=1
Max {2, 4} + 1 = 4 +1 = 5
Queue 2 Packet 1: t = 1, F(2,1) = R(1) + 3 = 1 + 3 = 4 2
L=3
Packet 2: t = 2, F(2,2) = max {F(2,1),R(2)} + 1 = 3
L=1 Max (4,2} + 1 = 5
Queue 3 Packet 1: t = 3, F(3,1) = R(3) + 5 = = 3+ 5 = 8 5
L=5

30