Bahria University, Islamabad Campus
Bahria University, Islamabad Campus
Bahria University, Islamabad Campus
Faculty’s Name: Dr. Muazzam A. Khan Max Marks: 50 Time Allowed: 90 Mins
Total Pages: 1 (including this)
INSTRUCTIONS:
I. All questions are compulsory.
II. There are total three questions. III.
The paper is closed book.
IV. The students are not allowed any helping material
(books, tables, formulas, etc). V. Use blue, black or blue-black
ink only. Do NOT use lead pencil especially.
Q.1 a. Briefly explain the common mistakes done in network performance evaluation?
Answer
Common Mistakes in Evaluation
1. No Goals
" No general purpose model
" Goals ⇒ Techniques, Metrics, Workload
" Not trivial
2. Biased Goals
" ``To show that OUR system is better than THEIRS'‘
" Analysts = Jury
3. Unsystematic Approach
4. Analysis Without Understanding the Problem
5. Incorrect Performance Metrics
6. Unrepresentative Workload
7. Wrong Evaluation Technique
8. Overlook Important Parameters
9. Ignore Significant Factors
10. Inappropriate Experimental Design
11. Inappropriate Level of Detail
12. No Analysis
13. Erroneous Analysis
14. No Sensitivity Analysis
15. Ignoring Errors in Input
b. What are different evaluation techniques you can use to evaluate the performance
of heterogeneous networks?
Answer
What are different evaluation techniques you can useto evaluate the performance of
heterogeneous networks?
Q.2 a. Explain the performance metrics you have chosen to evaluate your network &
what is the reason behind this selection?
Answer
When you’re evaluating your network performance, there are several different
metrics that you simply can analyze. Network performance is stricken by variety
of various factors. employing a network performance monitoring (NPM) solution,
your enterprise can rummage around for these factors and understand how they’re
hurting your network’s performance. the information your network monitoring
solution collects allows your IT team to realize insights into performance metrics
also.
It’s important for companies to understand which network performance metrics are
important to look at, though. Any metric will provide some information about
your network, which can make it appear to be you must analyze every single one.
However, counting on the particular issues that plague your network, not every
metric goes to be important for you to appear at. Despite this, there are some
metrics that are essential for any businesses to contemplate. Read on to get seven
essential network performance metrics that your company absolutely must keep
track of.
Bandwidth usage
Bandwidth is that the maximum data transmission rate possible on a network. For optimal
network operations, you would like to urge as near your maximum bandwidth as
possible without reaching critical levels. this means that your network is sending
the maximum amount data because it can within a period of your time, but isn’t
being overloaded. An NPM can monitor what quantity bandwidth is currently
being employed on a network, additionally as what quantity bandwidth is often
used during daily operations. the answer may provide you with a warning when
your network is using an excessive amount of bandwidth.
Throughput
Throughput measures your network’s actual data transmission rate, which might vary
wildly through different areas of your network. While your network’s bandwidth
measures the theoretical limit of knowledge transfer, throughput tells you ways
much data is really being sent. Specifically, throughput measures the proportion
of knowledge packets that are successfully being sent; a coffee throughput means
there are lots of failed or dropped packets that require to be sent again.
Latency
Latency is that the delay that happens between a node or device requesting data and when
that data is finished being delivered. This delay can happen for a spread of
reasons, but regardless of the cause, your NPM solution can track any delays and
log them. Consistent delays or odd spikes in delay time indicate a serious
performance issue; however, because delays can often be undetectable to the
human eye, you would like a monitoring tool to stay a watch on any delays that
happen.
Packet Loss
Packet loss examines what percentage data packets are dropped during data transmissions
on your network. The more data packets that are lost, the longer it takes for a
knowledge request to be fulfilled. Your IT team should understand how many
packets are being dropped on the average across your infrastructure. A network’s
Transmission Control Protocol (TCP) interprets when packets are dropped and
takes steps to make sure that data packets can still be transmitted; your network
team should monitor this method to create sure it’s working.
Retransmission
When packets are lost, the network must retransmit it to finish an information request. This
retransmission rate lets your enterprise understand how often packets are being
dropped, which is a sign of congestion on your network. you'll analyze
retransmission delay, or the time it takes for a dropped packet to be retransmitted,
to know how long it takes your network to live through packet loss.
Availability
Network availability, also know as uptime, simply measures whether or not the network is
currently operational. you'll never guarantee 100% availability, but you wish to
remember of any downtime that happens on your network that you just weren’t
expecting. It’s important to be alerted when the network goes down, which
network monitoring tools will provide for you. However, you must even be able
to discover your actual uptime percentage and the way often your network goes
down.
Connectivity
Connectivity refers as to if or not the connections between the nodes on your network are
working properly. If there's an improper or malfunctioning connection on your
network, it will be a serious hurdle for your company. Ideally, every connection
should be operating at peak levels in the least times. However, performance issues
like malware can target specific nodes or connections to affect performance in
this specific area of the network.
b. How you can configure a wireless node in NS-2, write down the commands
you can use in NS-2 tcl file for?
1. Create and config node
2. Trace file creation and tracing events
3. Create links between nodes
Answer
b. How you can configure a wireless node in NS-2, write down the commands
you can use in NS-2 tcl file for?
Ans:
The wireless networking model can be created using Tool Command Language (TCL)
script with fixed number of nodes. The sample code discussed below models the wireless
network with 2 nodes. Nodes are configured with the components of channel, networking
interface, radio propagation model, Medium Access Control (MAC) protocol, adhoc
routing protocol, interface queue, link layer, topography object, and antenna type. The
wireless network with 2 nodes can be viewed in the Network Animator (NAM) window
after executing the file sample1.tc
To configure a wireless node in NS-2
set ns_ [new Simulator] ;# Create a NS simulator object
$ns_ node-config \
-llType LL
-ifqType "Queue/DropTail/PriQueue"
-ifqLen 50
-macType Mac/802_11
-phyType "Phy/WirelessPhy"
-agentTrace ON or OFF
-routerTrace ON or OFF
-macTrace ON or OFF
-movementTrace ON or OFF
2. Trace file creation and tracing events
Answer
Techniques for workload Characterization
• Averaging
• Specifying dispersion
• Markov Models
• Single-parameter histograms
• Multi-parameter histograms
• Principal-component analysis
• Clustering
Averaging
• Specifying dispersion
•
• Variance and Standard deviation
• C.O.V
• Min-Max Range
• 10- 90-percentiles
• If C.O.V. is 0, then mean is the same
• If C.O.V. is high, may want complete histogram (next)
• Or divide into sub-components and average those that are similar only
• Ex: average small and large packet sizes
Clustering
Steps in Clustering
o Take Samples
o Select Parameters
o Transform
o Remove outliers
o Scale observation
o Select distant metrics
o Perform Clustering
o Interpret
Markov Models
If next request depends upon previous request, then can use Markov model
Actually, more general. If next state depends upon current state
If next request depends upon previous request, then can use Markov model
Actually, more general. If next state depends upon current state
Can use for application transitions
Example: users run editors, compilers, linkers
Markov model to characterize probability of type j after type i
Can use for page reference locality
Example: probability of referencing page (or procedure )But not probability
really refers to order of requests
May be several Markov models that have same relative frequency
c) Difference between Scatter Plots, Gantt Charts and Kiviat Graphs used for data
presentation?
Answer
Scatter Plots
Gantt charts
Kiviat Graphs
HB Metrics LB Metrics
App throughput App response time
Link utilization Link overhead
Router utilization Router overhead
% packets arrive # duplicates
% implicit acks % packets with error
d) What are different hardware and software tools used for network monitoring?
Answer:
Network Monitoring
Cacti. ...
Zabbix. ...
ntop. ...
Icinga. ...
Spiceworks. ...
Observium Community. ...
Wireshark.
The End of Question Sheet
Page 1 of 1