0% found this document useful (0 votes)
60 views35 pages

Nps Sprevidi 201102 Apricot

The document discusses how a neutral network measurement platform can benefit internet service providers, users, and authorities. It proposes that a neutral platform could empower users to monitor service quality, provide incentives for providers to collaborate on improving service, and serve as a watchdog for telecom markets. The document presents case studies showing how measurement data could help compare provider performance, enforce service level agreements, optimize network routes, identify inter-network problems, and detect major outages.

Uploaded by

ThanhNN0312
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
60 views35 pages

Nps Sprevidi 201102 Apricot

The document discusses how a neutral network measurement platform can benefit internet service providers, users, and authorities. It proposes that a neutral platform could empower users to monitor service quality, provide incentives for providers to collaborate on improving service, and serve as a watchdog for telecom markets. The document presents case studies showing how measurement data could help compare provider performance, enforce service level agreements, optimize network routes, identify inter-network problems, and detect major outages.

Uploaded by

ThanhNN0312
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 35

Net Measurement Neutrality:

An Enabler for Watchdog or Partnership?

Waiting Fok, Edmond Chan, Weichao Li,


Xiapu Luo, and Rocky Chang
Department of Computing
The Hong Kong Polytechnic University

1
A Net Measurement Neutral : Watchdog or Partner?

The (old) problem statement


• Information asymmetry between Internet
service providers and users
• Providers do not collaborate beyond peering.
• Possible results:
– Lack of incentives to innovate
– Competition based on competitive pricing
– Slow adoption of advanced technology

APRICOT 2011 2
A Net Measurement Neutral : Watchdog or Partner?

Possible solutions
• Empower users with the ability to monitor the
quality of service afforded to them.
– E.g., speed tests
• Empower users with the ability to monitor all the
providers’ quality of service.
– E.g., NetDiff
• Provide incentives for the providers to
collaborate on improving their quality of service.
– E.g., through a neutral net measurement platform

APRICOT 2011 3
A Net Measurement Neutral : Watchdog or Partner?

APRICOT 2011 4
A Net Measurement Neutral : Watchdog or Partner?

Is WIN-WIN-WIN possible?
• For providers
– Compare providers’ performance
– Enforce SLAs
– Troubleshoot network problems
• For users
– Independent value-for-money comparison
• For authorities
– Watchdog of telecom markets and services

APRICOT 2011 5
A Net Measurement Neutral : Watchdog or Partner?

Incentives for providers to participate


• Access to other providers’ performance data
which, however, do not reveal the providers’
business secrets
• A more accurate and timely diagnosis of
network problems
• Benchmark their service quality relative to
others’ service quality.
• Improve their service quality (and revenue).

APRICOT 2011 6
A Net Measurement Neutral : Watchdog or Partner?

Outline
• Our experience
• Case studies
– User’s perspective
• Compare ISPs’ performance
• SLAs enforcement
– ISP’s perspective
• Different uplinks selected
• Inter-network measurement
– Authority’s perspective
• Major network outages

APRICOT 2011 7
A Net Measurement Neutral : Watchdog or Partner?

Our experience with HARNET


• Continuous measurement since Jan 2009
• Probes at eight universities in Hong Kong
– 8Us share Internet connectivity through HARNET
– HARNET changed ISP in March 2010
– Monitor e2e paths to the same set of > 40 targets
• Collaboration with a major data center since 2007
• Collaboration with a major ISP since mid-2010
• Report findings to users and obtain feedback
• Visualize results to users in real time

APRICOT 2011 8
A Net Measurement Neutral : Watchdog or Partner?

APRICOT 2011 9
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance
• User select service provider based on price
and quality of service
• Service provider transition for HARNET during
24 Feb 2010 14:00 UTC to 7 Mar 2010 2:00
UTC
• ISP1  A temporary network  ISP2

APRICOT 2011 10
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance
Forward-path loss rates

APRICOT 2011 11
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance
Reverse-path loss rates

APRICOT 2011 12
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance
• Change of ISP and performance in HARNET
• Two-month periods immediately before and
after change-over from ISP1 to ISP2

FWL / RVL = Forward / Reverse-path packet loss


APRICOT 2011 13
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance

Performance for
Japan
RTT: ISP2 > ISP1

FWL: ISP2 > ISP1

RVL: ISP1 > ISP2

ISP2 
APRICOT 2011 14
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance

Performance for
the States
RTT: ISP1 > ISP2

FWL: ISP2 > ISP1

RVL: ISP1 > ISP2

ISP1 

APRICOT 2011 15
A Net Measurement Neutral : Watchdog or Partner?

Case study 1
Comparing two ISPs’ performance

Each ISP is good for


some (but not all)
paths.

APRICOT 2011 16
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement

APRICOT 2011 17
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement

APRICOT 2011 18
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement

APRICOT 2011 19
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement at HARNET

APRICOT 2011 20
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement at HARNET

APRICOT 2011 21
A Net Measurement Neutral : Watchdog or Partner?

Case study 2
SLA measurement at HARNET

APRICOT 2011 22
A Net Measurement Neutral : Watchdog or Partner?

Case study 3
Uplink selection
• Uplink selection based on the cost and quality
• Measuring the quality of the current routes
• Beyond load balancing

From Peplink

APRICOT 2011 23
A Net Measurement Neutral : Watchdog or Partner?

Case study 3
Uplink selection
• One of the links give poor performance
• Collaboration helps make the best selection.

APRICOT 2011 24
A Net Measurement Neutral : Watchdog or Partner?

Case study 3
Uplink selection

Changed back to
Changed to the shorter path
longer path

APRICOT 2011 25
A Net Measurement Neutral : Watchdog or Partner?

Case study 4
Inter-network measurement
• Active measurement reveals
– Forward-path traceroute
– Quality of e2e paths from a single source
• Unanswered questions:
– Reverse path?
– Problems inside or outside our network?

APRICOT 2011 26
A Net Measurement Neutral : Watchdog or Partner?

Case study 4
Inter-network measurement
• Co-operative two-way measurement
• Reveal both forward and reverse paths.

APRICOT 2011 27
A Net Measurement Neutral : Watchdog or Partner?

Case study 4
Inter-network measurement
• Significant asymmetric path for UB
– Much longer incoming paths
– Occurred during the ISP switchover
• UB prober observed dramatic increase in
delays to other Us, but no change the in
forward paths
– Long reverse paths

APRICOT 2011 28
A Net Measurement Neutral : Watchdog or Partner?

Case study 4
Inter-network measurement
(a) Affected paths

UH -> UB UB -> UH

UD -> UB UB -> UD

(b) Unaffected paths

UF -> UB UB -> UF

APRICOT 2011 29
A Net Measurement Neutral : Watchdog or Partner?

Case study 5
Major network outages
• Network outages cause huge financial loss.
• Damage of natural disasters are monitored by
Authorities:
– e.g. typhoon, earthquake, Internet service …
• Passively relying on ISP to report incidents
leads to slower responses.
• Collaborative active monitoring helps discover
the root cause and bypass the fault.

APRICOT 2011 30
A Net Measurement Neutral : Watchdog or Partner?

Case study 5
Major network outages
• Submarine optical fiber cables serving most
intercontinental traffic
• Outage of major cables  Reduction in the
international bandwidth
• Latest major event happened to Africa during
World Cup 2010
– One out of the two cables serving South Africa
was down.

APRICOT 2011 31
A Net Measurement Neutral : Watchdog or Partner?

Case study 5
Major network outage
• A SEACOM cable fault occurred at 9:19 on 5
July 2010.
• Significant traffic re-routed to SAT-3
– Bandwidth reserved for TV broadcasting
– Congestion, packet loss (> 20%), high latency (1sec)
• Many companies and home users in South
Africa lost International connectivity.

APRICOT 2011 32
A Net Measurement Neutral : Watchdog or Partner?

Case study 5
Major network outage
• Collaborative monitoring immediately raises
alerts based on network quality measurement.

World Cup Final


ends Repair completed
SEACOM cable on 23 July
fault at 9:19 5 July

33
APRICOT 2011
A Net Measurement Neutral : Watchdog or Partner?

To conclude …
• A neutral net measurement platform provides
trustworthy information.
– Benefit users
• Get the best combination of cost and service quality.
• Ensure that ISPs maintain their service standards.
– Co-operative network monitoring helps ISPs
• Make better business / routing decision.
• Troubleshoot problem.
– Necessary for authorities
• Quick response to incidents

APRICOT 2011 34
Net Measurement Neutrality
A Watchdog: promote a fair market
and ensure reliable service
Your Partner: encourage innovation
and improve service quality

35

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy