0% found this document useful (0 votes)
92 views

Network Performance

The document discusses criteria for evaluating networks, including performance, reliability, and security. It describes how performance is measured in terms of delay, throughput, bandwidth-delay product, and jitter. Reliability is measured by failure and recovery times using MTBF and MTTR metrics, which are used to calculate availability. Security involves protecting data from errors and malicious users.

Uploaded by

Justine Joy
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
92 views

Network Performance

The document discusses criteria for evaluating networks, including performance, reliability, and security. It describes how performance is measured in terms of delay, throughput, bandwidth-delay product, and jitter. Reliability is measured by failure and recovery times using MTBF and MTTR metrics, which are used to calculate availability. Security involves protecting data from errors and malicious users.

Uploaded by

Justine Joy
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

Criteria for Evaluating Networks

 Performance
 Measured in terms of Delay, Throughput, Bandwidth-delay product, and Jitter
 Quality of the network in terms of packet loss and congestion
• Affects greatly the throughput and delay
 Reliability
 Failure and recovery (MTBF / MTTR)
 Measured in terms of availability/robustness
 Security
 Data protection against corruption/loss of data due to:
 Errors
 Malicious users

1
Network Performance
Throughput: Amount of data that can Latency = propagation time +
pass through the network transmission time + queuing time +
processing delay
Bandwidth-delay product: number Propagation time = distance /
of bits that can fill the link propagation speed
Transmission time = data size /
bandwidth
Queuing time = time spent in the
Latency or Delay: Length of time it interface buffer
takes for an entire message to completely Processing delay = Processing time
arrive at the destination from the time the
Jitter - the variation in delay for
first bit is sent out from the source packets belonging to the same flow
2
B. Forouzan, Chp 3, Data Communications and Networking: Fifth Edition. McGraw-Hill Higher Education, pp 84-89, 2012
Network Performance
Throughput computation an example
What is the throughput of a network that can send 1000
frames/sec carrying approximately 10,000 bits per frame? What
is the efficiency of this network if its bandwidth is 100 Mbps?
Throughput ________________ Mbps
Efficiency ________________ %

3
B. Forouzan, Chp 3, Data Communications and Networking: Fifth Edition. McGraw-Hill Higher Education, pp 84-89, 2012
Throughput
= 10,000 bits / frame x 1000 frame / sec
= 10,000,000 bits/sec
= 10 Mbps
Efficiency
= actual utilization / bandwidth x 100 %
= 10 Mbps / 100 Mbps x 100 %
= 10 %

10 % utilization, 90% under-utilized


Recommended: 60 to 80 % utilization; you have 20 to 40% for
burst traffic and control packets
Not 100% utilization of the capacity; implies congestion
Latency, an Example Computation
Latency = propagation time + What is the total delay (latency) for a
transmission time + queuing time + frame of size 5 million bits that is being
processing delay sent on a link with 10 routers each
having a queuing time of 2 μs and a
Propagation time = distance /
processing time of 1 μs.
propagation speed
The length of the link is 2000 Km. The
Transmission time = data size / speed of signal through the link is 2 ×
bandwidth 108 m/s.
Queuing time = time spent in the The link has a bandwidth of 5 Mbps.
interface buffer
Which component of the total delay is
Processing delay = Processing time dominant? Which one is negligible?
6
B. Forouzan, Chp 3, Data Communications and Networking: Fifth Edition. McGraw-Hill Higher Education, pp 84-89, 2012
Latency / Delay
What is the total delay (latency) for a Propagation time = 2,000,000m /
frame of size 5 million bits that is 200,000,000 m/s = .01 sec
being sent on a link with 10 routers
each having a queuing time of 2 μs Transmission time
and a processing time of 1 μs.
= 5 Mbits/ 5Mbps = 1 sec
The length of the link is 2000 Km. The
speed of light inside the link is 2 × 108 Queuing time = 10*2 μs = 10
m/s. * .000002 = .000020 = .00002 sec
The link has a bandwidth of 5 Mbps. Processing delay = 10 * 1 μs = 10
Which component of the total delay *.000001 = .000010 = .00001 sec
is dominant? Which one is negligible? Latency = 1.01003 sec
7
B. Forouzan, Chp 3, Data Communications and Networking: Fifth Edition. McGraw-Hill Higher Education, pp 84-89, 2012
Bandwidth-Delay Product

Bandwidth-delay product


A link with a bandwidth of 1 bps has a delay of 5 secs

What is the bandwidth-delay product

Explain its meaning

Another link has a bandwidth of 5 bps with a delay of 5 secs

What is the bandwidth-delay product

Explain its meaning
8
Jitter
B. Forouzan, Chp 28, Data Communications and Networking: Fifth Edition. McGraw-Hill Higher Education, p 991, 2012

https://howdoesinternetwork.com/2013/jitter
9
( xi − x) 2 Jitter
Jitter = ∑ n−1 how far the packet inter-arrival time
is spread out from their average value
x1 = 11-1 = 1 ; x2 = 21-11 = 1 ;
x3 = 31-21 = 1 ; n =3
x 1 + x2 + x 3 1+1+1
x= = =1
3 3
2 2 2
Jitter = (1−1) +(1−1) +(1−1) =0
3
( xi − x) 2 Jitter
Jitter = ∑ n−1 how far the packet inter-arrival time
is spread out from their average value
x1 = 15-1 = 14 ; x2 = 27-15 = 12 ;
x3 = 37-27 = 10 ; n =3
x 1 + x2 + x 3 14+12+10
x= = =12
3 3
2 2 2
Jitter = (14−12) +(12−12) +(10−12) = 4+0+4 =4
3−1 2
Criteria for Evaluating Networks
 Performance
 Measured in terms of Delay, Throughput, Bandwidth-delay product, and Jitter
 Quality of the network in terms of packet loss and congestion
• Affects greatly the throughput and delay
 Reliability
 Failure and recovery (MTBF / MTTR)
 Measured in terms of availability/robustness
 Security
 Data protection against corruption/loss of data due to:
 Errors
 Malicious users

12
How robust is the network?
Failure and Recovery

A given network was deployed and used for one (1) year

In its logbook was written that it was down in February
and was repaired for 4 days. Then, it went down again
in May and November and was repaired for 3 days and
1 day respectively

What is the mean time before failure (MTBF), mean time
to repair (MTTR), and Availability
13
How robust is the network?
Failure and Recovery
MTBF=(operational time)/(number of downtime)
= (365 – 4 – 3 – 1)/3 = 357/3 = 119 days
MTTR=total downtime / number of repairs = (4+3+1)/3=2.67
Availability = MTBF / (MTBF + MTTR)
= uptime / operational time
= 357 / 365 = 97.81%
14
How reliable is the network?
Parallel Systems: A = 1 – (1-AA) * (1-AB)
= 1 – (1-0.8)*(1-0.8) System A Component A
= 1 – 0.2*0.2 Availability = 0.8
= 0.96 Component B
Availability = 0.8

Series Systems: A = AA * AB

System B Component A Component B


=AA*AB Availability = 0.8 Availability = 0.8
= 0.8*0.8
= 0.64
15
How reliable is the network?
Hybrid Systems: System A:
AAB = 1–(1–0.7)x(1–0.8 ) = 1–0.06
System A = 0.94
Component A Component C
Availability = 0.7 Availability = 0.9
ACD = 1–(1–0.9)x(1–0.6 ) = 1–0.04
= 0.96
Component B Component D AABCD = 0.96*0.94
Availability = 0.8 Availability = 0.6
= 0.9024 or 90.24%
System B:
System B AAC = 0.7*0.9 = 0.63
Component A Component C
Availability = 0.7 Availability = 0.9 ABD = 0.8*0.6 = 0.48
AACBD = 1–(1–0.63)x(1–0.48 ) = 1 – 0.37x0.52
Component B Component D
Availability = 0.8 Availability = 0.6
= 1–0.1924
= 0.8076 or 80.76%

16
Thank you!

17

You might also like