ITU REGIONAL STANDARDIZATION FORUM
FOR AFRICA
(Dakar, Senegal, 23-24 March 2015)
Tools and Methods for Testing the QoS
of Internet services
Yvonne Umutoni
Yvonne.umutoni@rura.rw
1
EXECUTIVE SUMMARY
• This presentation focuses on the quality of
service testing tools and methodologies of
internet services.
• It covers the following:
– QoS Offered and QoS Achieved
– Four view points of QoS according to ITU-T E.800
– Testing Methods
– Testing Tools
– Sampling Methodology
2
o Why is it needed to test the QoS of Internet
services offered by ISPs?
o What are the measurement methodologies and
tools that can be adopted in order to evaluate the
QoS of Internet delivered by ISPs to the
customers versus the offered/promised?
3
QOS OFFERED AND QOS ACHIEVED
The QoS of Internet offered (advertised) by ISPs is
tested:
o by ISPs for optimization purposes or
o by Regulators because sometimes the QoS
promised (in advertisements) is totally different
with the QoS Delivered/achieved by ISPs.
4
DIFFERENT ASPECTS OF QOS
Customer Service
Customer’s QoS Provider
QoS Offered by
requiremen Service
ts Providers
QoS
End-User Achieved
Perception by Service
Providers
Four view points of QoS according to ITU-T E.800
5
TESTING METHODS
o Method 1:Measurements/testing using injected
traffic
o Method 2: Measurements/testing using user
traffic
o This presentation focuses on method 1
6
METHOD 1: TESTING USING INJECTED
TRAFFIC
o Performs analysis based on sending traffic (probing
packets) between two destinations;
o Probing packets are injected in the network connection to
measure the quality of service (QoS) of different services
(web browsing, file transfer, VoIP, etc) over Internet
connections.
o There are three scenarios for method 1:
o Test at the ISP level,
o Test at the National level (at Local IXP) and
o Test at International level (at International IXP)
7
1ST SCENARIO: TEST AT ISP LEVEL
o Test server is
located to the edge
of the network
(peering router)
o More used to
evaluate internet
bottlenecks inside
ISP’s network
o Relevant to the ISPs o This approach can not be used to know the
customer experience when accessing the internet
because the information measured refers to the
ISP network
8
2ND SCENARIO: TEST AT NATIONAL LEVEL
o Test server is located a
the local Internet
exchange point
o Can be carried on with
or without involvement
of ISPs by Regulators
o Measurement path
includes a complete
Internet connection from
customer to test server
located at the local IXP
o This scenario allows the o Applied when testing local download/upload speed, latency
benchmarking of the
QoS access of different o The benchmarking/comparability of the ISPs can be reached
ISPs to the local Internet in the best way, in case all ISPs are connected in a similar
exchange point. way to that local IXP (or any central measuring point)
9
2ND SCENARIO: TEST AT NATIONAL LEVEL
(Cont…)
• In the case of web browsing download speed at
national level, the measurements are done on
the selected web pages stored on servers
located inside the country.
10
3RD SCENARIO: TEST AT INTERNATIONAL LEVEL
o Test server located to
International Internet
exchange point ( IXP of
another country, may be
another continent)
o Usually, the Internet
connection that ISPs
provide to customers is to
the entire internet.
o Therefore, the more
bandwidth capacity in the o Applied when testing international
ISPs’ connections , the download/upload speed
more better the quality of o This allows the comparison between the
Internet connection connectivity of internet services inside the country
provided by ISPs. and outside ( different countries/ continents)
11
3RD SCENARIO: TEST AT INTERNATIONAL
LEVEL (Cont…)
• In the case of web browsing download speed
internationally, the measurements are done on
the selected web pages stored on servers
located outside the country.
12
TESTING TOOLS
• Hardware-based Tools (Probes) and
• Software-based Tools
13
HARDWARE BASED TOOLS
There are two options of implementation:
o The probe completely replaces the end user’s
equipment.
o No other equipment can be connected to the Internet access
while the probe is performing measurements.
o This is applicable also in the case of mobile Internet access.
o The probe shares the Internet access with ordinary
traffic
o For example by connecting a probe to a customer’s
residential gateway.
14
SOFTWARE BASED TOOLS
There are two kind of software based tools :
o A web-based tool: where the download and execution
of the measurement software is initiated via the end
user’s web browser by accessing a specific web page.
o A dedicated software client: where the measurement
software is permanently installed on the end user’s
terminal equipment.
o In this case, different versions of the software are needed to
support different operating systems and terminal equipment.
15
SAMPLING METHODOLOGY
o Selection of panelists (end user access points, where to install
probes) should take into consideration various factors such as
technologies, Internet speed packages (depending on
popularity), locations and so on.
o identification of panelists based on statistical sampling:
o With statistical sampling theory, we need to identify the dependent
variable (e.g: download/upload speed/throughput)
o Then, you look at other influencing factors (explanatory variables) to the
dependent variable.
o For example the primary explanatory variable in case of a country could
be location while the secondary explanatory variable could be the
distance between the probe and the test server location.
16
THANK YOU FOR YOUR
ATTENTION!
17
