measuring network performance Flashcards
what do we mean by the internet is a “best effort” delivery service
some packets wont get delivered
the can get reordered or delayed
there is no guarantee on the condition theyll arrive in
store and forward
the entire packet must arrive at the router before it can be transferred on the next link
where does reliability over the internet come from
protocols
what is a delay and how is it measured
the time taken for a packet to travel across the network
measured in fractions of a second
transmission delay
the time it takes to push all the bits of the packet into the link
bits have to be pushed one at a time
how do you calculate the transmission delay
packet length(bits) / link transmission rate(bps)
queuing delay
when work travels faster than it can be processed packets have to be stored in the router buffer until the router is ready to transmit them
packet losses
when the packets in the router buffers reach capacity, the router starts to discard them
in which other ways can packet losses occur
packets can get lost in the physical medium
hardware and software malfunctions; errors, corruption, failed checksums
attacks
what are the 4 types of delay
transmission delay
queuing delay
processing delay
propagation delay
what are the 3 ways that the router buffers choose which packets to drop
tail end drop
random drop
quality of service (qos) aware (priority)
random drop
buffer drops any packet in the queue
qos aware
packets will be dropped according to their priority
what is the benefit of qos aware drop
provides fairness
guaranteed throughput for sensitive services e.g. voice calls and live videos
how do dos attacks cause packet losses
bombarding a device with packets so its buffers are filled up which can prevent other hosts from sending packets
processing delay
the time taken for a device to examine the packets header and decide where to direct it
usually how long is the processing delay
usually microseconds bit can vary depending on how busy the device is
propagation delay
time needed to travel from one end of the link to another
what is propagation delay dependant on
the physical type of the link
how do you calculate the propagation delay
physical link length / link’s propagation speed
what is the difference between the propagation speed for wired and wireless links
wired; approx 2/3 the speed of light
wireless; v close to the speed of light
nodal delay
the total of all the different types of delays measured per node
throughput
the rate in bits/s at which bits are transferred from a sender to a receiver
end-to-end delay
total of all nodal delays from one end to another
round- trip time (rtt)
end to end delay measured in both directions
from one end to another and back but doesn’t have to be the same route
what does the traceroute program do (not how)
it measures delay from the source to each router on the path to the destination and records the rtt for each hop
during traceroute, for each router, what does the source do
sends 3 packets/probes to a router that lies on the path to the destination
the router returns the packets to the sender
the sender measures the lag between the transmission and the reply
what does *** mean in tracerotue
no response; router not responding / probe lost
what do we use to measure delay
traceroute
how do we calculate the traffic intensity (queuing delay)
(l * a) / r = arrival rate of bits / service rate of bits
instantaneous throughput
the rate at a given point in time
average throughput
rate over a period of time
peak throughput
highest instantaneous throughput rate seen so far
bottleneck link
link on the end-to-end path that constraints end-to-end throughput
what does it mean if traffic intensity is ~ 0
average delay is small
what does it mean if traffic intensity is -> 1
average delay is large
what does it mean if traffic intensity is > 1
more work is arriving than can be processed
average delay is infinite
what is usually the bottleneck link
Rc (network capacity) or Rs (sending rate)
what do we use to measure throughput
using iperf; provides standardized throughput measures
how does iperf normally run (model)
as a client server model
how does iperf work
client host requests sample data and the server host serves it which generates a stream of sample data across the network
because we know how large the sample data is and how long it takes to retrieve it we can calculate throughput
what are some benefits of iperf
v configurable; can use different protocols and data sample sizes
can do bidirectional transfer to test both directions and check for inequalities
how do we measure loss
ping
how does ping work
creates messages to be sent out to specific hosts
measures the rtt
measures packet loss by keeping track of how many messages were sent and how many responses were received
