14
Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the International Development Research Centre, Canada and the Department for International Development, UK.

Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Embed Size (px)

Citation preview

Page 1: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Using Volunteer Computing for monitoring Broadband QoSE

Chanuka Wattegama and Nilusha Kapugama

This work was carried out with the aid of a grant from the International Development Research Centre, Canada and the Department for International Development, UK.

Page 2: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Agenda

• Monitoring Quality of Service Experience (QoSE) : possible approaches

• Monitoring QoSE: through demand-side testing• Volunteer Computing for data gathering• Results: from USA and India• Conclusion

Page 3: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

We measure QoSE: from the demand (user) side, objective and quantifiable

QoSE: Quality of Service Experience

• An objective measure of user’s experience of delivered quality with or without reference to promised levels.

Quality of ExperienceQuality of

Service

Page 4: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Possible ways to monitoring quality Self

regulation by operators

Direct monitoring

by regulators

User satisfaction

surveys

Demand-side (user) testing

Intrusiveness on network

+ +++ - +

Regulator participation

++/+/- +++ Depends on survey -

Operator participation

+++ +++ Depends on survey -

User participation

- - +++ +++

Subjectivity of results

++/+/- + +++ +

+++ High ++ Medium + low - none

Page 5: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Relevant metrics for testing

+++ Highly Relevant ++ very relevant + relevant - not relevant

Throughput Delay

Service Down Up RTT Jitter Loss

Browse (text) ++ - + - -

Browse (media) +++ - + + +

Download file +++ - - - -

Upload File - +++ - - -

Transactions + + ++ + +

Streaming media +++ - + ++ ++

VOIP + + +++ +++ +++

Games ++ + +++ ++ ++

Page 6: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Available Software Applications monitor…

Metric Speedtest.net

Speedtest2.com

internetfrog

Download

Upload

RTT

Jitter

Packet Loss

Availability

AT-Tester

Application Developed by IIT-Madras uses 6 Metrics

Page 7: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Methodology contd…

• Software freely available for download• Multiple measures of each metric taken during a day

– The time and day is recorded and reported to check for Peak/Off peak variations

• Location is recorded to measure regional variations• The metrics are measured in three domains; ISP,

National, International

Page 8: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

8

66.163.169.xxx

InternetInternet

User

ISP (Sri Lanka)

Singtel (Singapore)

Network Solutions(USA)

Yahoo220.247.232.xxx 222.165.130.xxx

203.208.151.xxx

National Server(Sri Lanka)

202.69.200.xxx

The 3 test domains…

Page 9: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Results on uploaded to www.broadbandasia.info

Page 10: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Methodology depends on a hybrid of Volunteer Computing to get sufficient data

Volunteer Computing characteristics AT-Tester

Uses storage and processing capacity of volunteers’ computersSoftware application available for download

Computation done by the application, results uploaded to websiteNo effort required from the volunteer – uses computer when idle ×Academic or public interest project

Trustworthy Software Application

Uses computer processing and storage resources of the general public (volunteers) for computationally intensive projects (Anderson, 2009)

Page 11: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Users in USA get more value for money

Download from National Domain Download from International Domain

0

20

40

60

80

100

120

800 1100 1500 1800 2000 2300Time

Kbps

per D

ollar

Comcast (6Mbps) - Denver Verizon (3 Mbps) BuffoloBSNL (256 Kbps) Chennai Airtel (256 Kbps) Delhi

0

20

40

60

80

100

800 1100 1500 1800 2000 2300Time

Kbps

per D

ollar

Comcast (6Mbps) - Denver Verizon (3 Mbps) BuffoloBSNL (256 Kbps) Chennai Airtel (256 Kbps) Delhi

Page 12: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

0

100

200

300

400

National International

Late

ncy

(ms)

Airtel (256 kbps) Delhi BSNL (256 kbps) Chennai Comcast (6Mbps) - Denver Verizon (3 Mbps) -Buffolo

Latency to national level similar but diverges at International level

– * Limits specified by the Singapore Regulator IDA for national and International network latency

300 ms*

85 ms*

Page 13: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

Does the model Work?

• Methodology has potential – but adoption levels need to be improved.

– Number of volunteers lower than expected – Many register/download software; but few complete

• Possible ways to increase adoption: – More publicity (More volunteers from Sri Lanka and India

where awareness work (workshops, public lectures) was done

– Incentives (free software for highest number of readings contributed etc)

– Seeing positive results (through operator and policy makers taking actions based on results)

Page 14: Using Volunteer Computing for monitoring Broadband QoSE Chanuka Wattegama and Nilusha Kapugama This work was carried out with the aid of a grant from the

For comments, questions and feedback

[email protected]@lirneasia.net