Advanced network performance measurement techniques

Martin Geddes
Martin GeddesComputer scientist and telecoms consultant at Martin Geddes Consulting Ltd.
Advanced network performance
measurement techniques
Dr Neil Davies
Predictable Network Solutions Ltd
Peter Thompson
Predictable Network Solutions Ltd
Martin Geddes
Martin Geddes Consulting Ltd
PREDICTABLE
NETWORK
SOLUTIONS
© 2013 All Rights Reserved
Dr Neil Davies
Co-founder, Predictable Network Solutions Ltd
Peter Thompson
CTO, Predictable Network Solutions Ltd
Martin Geddes
Founder, Martin Geddes Consulting Ltd
PREDICTABLE
NETWORK
SOLUTIONS
The only ex ante network performance
engineering company in the world.
Consultancy on the future of
telecoms and the Internet.
PREDICTABLE
NETWORK
SOLUTIONS
Context for this presentation
We are all in the business of “information translocation”
The timely movement of information from
one computational process to another
The value lies in delivering application outcomes
That people will pay for
You are reading this because you are interested in
delivering successful outcomes
And understanding the causes of failure, so they can be mitigated
You may be working in a culture of deflecting
the attribution of blame
We’d like to help you turn away from the path to the Dark Side
What affects the timeliness?
• The timeliness of application outcomes is dependent on the
end-to-end loss and delay characteristics of the
translocation
• We call this end-to-end property ∆Q
– ∆Q applies in each direction – not just the round trip
– These characteristics need to be suitably bounded
• ∆Q depends on the offered load
– “Bandwidth” is an aspect of the relationship between offered
load and ∆Q
This presentation is about measuring ∆Q
– and the benefits that approach brings
Good measurement is
NOT about averages
• The average number of legs of a Swedish person is 1.9
– Now find me one!
• Measuring average throughput on a 1Gb link over 10
mins is like measuring the traffic on the M5 motorway
over two years
– No indicator of my likely travel experience
• Need to know the instantaneous properties
– The ∆Q the “next packet” is going to get
• It is all about the probability distribution of quality
attenuation
– This is what determines timeliness of application outcomes
One-point measures
• This is the typical information captured by equipment
today
– Counters (e.g. packets passed, packet sizes, packets
dropped)
– Sampled over a period
• Does not capture ∆Q
– Not end-to-end
• Multiple one-point measures don’t help
• Creates an equipment-centric view
– Focuses on the equipment, not the service to the
customers
– Leads to focus on capacity, and ignores schedulability
Multipoint measures
• Measure a value between different points
– Not just counting things
• Same “information translocation” at various
points
– Measuring the dynamics of the flow
• Isolates issues, in both space and time
– Excellent diagnostic power
• Leads to a focus on schedulability and trading
– Which in turn focuses on the outcomes for the
customer
Different measurement approaches
Average Instantaneous
Single
Point
Offered Load
and Utilisation
(mean values only)
Limited predictive power
Arrival Patterns
Temporal predictive power,
localised assurance
(compliance with
arrival pattern policy)
Multiple
Point
Delay and Loss
(mean and variance)
Spatial predictive power
Temporal and spatial
predictive power
Assurance of both arrival and
service (demand and supply) –
represents all that can be known
about a system (by observation)
PLUS
PLUS
Interpreting the two-point measure
© Predictable Network Solutions 2013
Raw data: There’s no discernable structure here
– not possible to work with data like this.
Sort by packet size:
a clear structure
emerges
Serialisation (or
size-related) delay
S
Packets with bigger payloads experience more structural delay:
it takes longer to turn the packet into a bitstream, and back again into a packet
at the next network element .
© Predictable Network Solutions 2013
Geographic delay
Serialisation delay
Variable contention
delay
G
S
V
Example multi-point measure
The bi-directional, end-to-end path of a small cell
deployment over commodity infrastructure
How to read the information
• Different views tell different stories
• We’ll see some of those stories in the
following slides
• The focus on V is because that is where the
issues of schedulability manifest themselves
Key to following charts
Two point
measures (by time)
GSV view
(by packet)
V (by time) V (by packet size)
V cumulative
distribution
function (main)
V cumulative
distribution
function (tail)
E to A direction (user experience)Return Transit (run dd0a2310-d235-495b-8d2f-a4dc
0
0.05
0.1
0.15
0.2
0 50 100 150 200 250
delay(s)
run time (s)
Observed Delay against Experiment Run Time
E->A
0
0.05
0.1
0.15
0.2
0
delay(s)
0.02
0.04
0.06
0.08
0.1
0.12
0.14
delay(s)
Observed Delay Variability (V) against Experiment Run Time
E->A
0.02
0.04
0.06
0.08
0.1
0.12
0.14
delay(s)
Note the delay spike during the test run @ approx 60 seconds in
How can be begin to analyse this performance issue?
E -> A (by packet)
This ‘spike’ doesn’t appear to be related to a particular packet size
(note ‘striations’ in the S value is an artefact of 3GPP scheduling)
E -> A (Dynamic response)
Removing G and S influences clearly highlights
the magnitude of the contention issue
Spatial Isolation
Same magnitude
issue between D
to B
But not between
D and C
Spatial Isolation (2)
It is occurring
between C and B
NOTE: this is the effect that we are measuring
– NOT the cause (which in this case was not the access
network but elsewhere)
Armed with this information, we can begin to analyse root
causes (e.g. what is over-driving this link?)
∆Q for ADSL line
© Predictable Network Solutions 2013
Compare and contrast: baseline data for an ADSL line
∆Q for Femto (over ADSL)
© Predictable Network Solutions 2013
Now run a femtocell over that same line: much worse performance
Summary
• Multipoint distribution based measurement gives
access to all the information available through
observation
– “observation” is key – independent of equipment
– Captures the influence of technology etc
• G,S & V gives you a way of extracting both
temporal and spatial details
• Becomes extremely powerful when combined
with analysis
– E.g. you have a model of what V should be, or what G
and S should be given the network layout
Upcoming workshops:
Sustainable Public Service Networks
London, 19th September 2013
Fundamentals of
Network Performance
London, 20th September 2013
www.sustainablebroadband.com
PREDICTABLE
NETWORK
SOLUTIONS
Neil Davies
Neil.Davies@pnsol.com
Peter Thompson
Peter.Thompson@pnsol.com
Martin Geddes
mail@martingeddes.com
PREDICTABLE
NETWORK
SOLUTIONS
1 of 26

Recommended

Why ∆Q is the ideal network metric by
Why ∆Q is the ideal network metricWhy ∆Q is the ideal network metric
Why ∆Q is the ideal network metricMartin Geddes
1.9K views27 slides
The Ladder: How money and multiplexing are connected by
The Ladder: How money and multiplexing are connectedThe Ladder: How money and multiplexing are connected
The Ladder: How money and multiplexing are connectedMartin Geddes
5.5K views120 slides
Geddes/PNSol - Broadband market evolution by
Geddes/PNSol - Broadband market evolutionGeddes/PNSol - Broadband market evolution
Geddes/PNSol - Broadband market evolutionMartin Geddes
7K views35 slides
Network performance optimisation using high-fidelity measures by
Network performance optimisation using high-fidelity measuresNetwork performance optimisation using high-fidelity measures
Network performance optimisation using high-fidelity measuresMartin Geddes
6.4K views27 slides
Introduction to ΔQ and Network Performance Science (extracts) by
Introduction to ΔQ and Network Performance Science (extracts)Introduction to ΔQ and Network Performance Science (extracts)
Introduction to ΔQ and Network Performance Science (extracts)Martin Geddes
3.2K views11 slides
Essential science for broadband regulation by
Essential science for broadband regulationEssential science for broadband regulation
Essential science for broadband regulationMartin Geddes
2.6K views77 slides

More Related Content

What's hot

FCC Open Internet Transparency - a review by Martin Geddes by
FCC Open Internet Transparency - a review by Martin GeddesFCC Open Internet Transparency - a review by Martin Geddes
FCC Open Internet Transparency - a review by Martin GeddesMartin Geddes
1K views39 slides
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’ by
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’Martin Geddes
2.5K views78 slides
Network Cost and Performance Transformation Services by
Network Cost and Performance Transformation ServicesNetwork Cost and Performance Transformation Services
Network Cost and Performance Transformation ServicesMartin Geddes
1.5K views13 slides
BT Operate Case Study by
BT Operate Case StudyBT Operate Case Study
BT Operate Case StudyMartin Geddes
930 views5 slides
What is a 'polyservice' network? by
What is a 'polyservice' network?What is a 'polyservice' network?
What is a 'polyservice' network?Martin Geddes
1.2K views9 slides
Stationarity is the new speed by
Stationarity is the new speedStationarity is the new speed
Stationarity is the new speedMartin Geddes
1.7K views49 slides

What's hot(20)

FCC Open Internet Transparency - a review by Martin Geddes by Martin Geddes
FCC Open Internet Transparency - a review by Martin GeddesFCC Open Internet Transparency - a review by Martin Geddes
FCC Open Internet Transparency - a review by Martin Geddes
Martin Geddes1K views
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’ by Martin Geddes
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’
Addicted to speed: Why broadband service providers need a ‘healthier lifestyle’
Martin Geddes2.5K views
Network Cost and Performance Transformation Services by Martin Geddes
Network Cost and Performance Transformation ServicesNetwork Cost and Performance Transformation Services
Network Cost and Performance Transformation Services
Martin Geddes1.5K views
What is a 'polyservice' network? by Martin Geddes
What is a 'polyservice' network?What is a 'polyservice' network?
What is a 'polyservice' network?
Martin Geddes1.2K views
Stationarity is the new speed by Martin Geddes
Stationarity is the new speedStationarity is the new speed
Stationarity is the new speed
Martin Geddes1.7K views
Introduction to network quality arbitrage by Martin Geddes
Introduction to network quality arbitrageIntroduction to network quality arbitrage
Introduction to network quality arbitrage
Martin Geddes748 views
Sample proposal summary for quality arbitrage business unit by Martin Geddes
Sample proposal summary for quality arbitrage business unitSample proposal summary for quality arbitrage business unit
Sample proposal summary for quality arbitrage business unit
Martin Geddes553 views
Network performance - skilled craft to hard science by Martin Geddes
Network performance - skilled craft to hard scienceNetwork performance - skilled craft to hard science
Network performance - skilled craft to hard science
Martin Geddes920 views
Connect, Interact, Transact by Martin Geddes
Connect, Interact, TransactConnect, Interact, Transact
Connect, Interact, Transact
Martin Geddes2.4K views
CARE initiative technical prospectus by Martin Geddes
CARE initiative technical prospectusCARE initiative technical prospectus
CARE initiative technical prospectus
Martin Geddes558 views
Digital supply chain quality management by Martin Geddes
Digital supply chain quality managementDigital supply chain quality management
Digital supply chain quality management
Martin Geddes549 views
The Properties and Mathematics of Data Transport Quality by Martin Geddes
The Properties and Mathematics of Data Transport QualityThe Properties and Mathematics of Data Transport Quality
The Properties and Mathematics of Data Transport Quality
Martin Geddes1.6K views
Navigating the Uncertain World Facing Service Providers - Juniper's Perspective by Juniper Networks
Navigating the Uncertain World Facing Service Providers - Juniper's PerspectiveNavigating the Uncertain World Facing Service Providers - Juniper's Perspective
Navigating the Uncertain World Facing Service Providers - Juniper's Perspective
Juniper Networks7.3K views
Performance and Supply Chain Management for the Software Telco by Martin Geddes
Performance and  Supply Chain Management for the Software TelcoPerformance and  Supply Chain Management for the Software Telco
Performance and Supply Chain Management for the Software Telco
Martin Geddes600 views
Study on reliability optimization problem of computer By Dharmendra Singh[Srm... by Dharmendrasingh417
Study on reliability optimization problem of computer By Dharmendra Singh[Srm...Study on reliability optimization problem of computer By Dharmendra Singh[Srm...
Study on reliability optimization problem of computer By Dharmendra Singh[Srm...
Advanced techno-economic modelling of distribution network investment require... by Power System Operation
Advanced techno-economic modelling of distribution network investment require...Advanced techno-economic modelling of distribution network investment require...
Advanced techno-economic modelling of distribution network investment require...

Similar to Advanced network performance measurement techniques

Future of Broadband workshop presentation - ITU Telecom World 2013 by
Future of Broadband workshop presentation - ITU Telecom World 2013Future of Broadband workshop presentation - ITU Telecom World 2013
Future of Broadband workshop presentation - ITU Telecom World 2013Martin Geddes
16.7K views85 slides
Smart Grids and Big Data by
Smart Grids and Big DataSmart Grids and Big Data
Smart Grids and Big DataDave Callaghan
556 views56 slides
Capgemini ses - smart grid operational services - where to start - five fou... by
Capgemini   ses - smart grid operational services - where to start - five fou...Capgemini   ses - smart grid operational services - where to start - five fou...
Capgemini ses - smart grid operational services - where to start - five fou...Gord Reynolds
217 views8 slides
Data Science, Knowledge Discover, Mining and Learning by
Data Science, Knowledge Discover, Mining and LearningData Science, Knowledge Discover, Mining and Learning
Data Science, Knowledge Discover, Mining and LearningEUBrasilCloudFORUM .
88 views10 slides
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N... by
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...IRJET Journal
3 views7 slides
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in... by
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...IRJET Journal
17 views2 slides

Similar to Advanced network performance measurement techniques(20)

Future of Broadband workshop presentation - ITU Telecom World 2013 by Martin Geddes
Future of Broadband workshop presentation - ITU Telecom World 2013Future of Broadband workshop presentation - ITU Telecom World 2013
Future of Broadband workshop presentation - ITU Telecom World 2013
Martin Geddes16.7K views
Capgemini ses - smart grid operational services - where to start - five fou... by Gord Reynolds
Capgemini   ses - smart grid operational services - where to start - five fou...Capgemini   ses - smart grid operational services - where to start - five fou...
Capgemini ses - smart grid operational services - where to start - five fou...
Gord Reynolds217 views
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N... by IRJET Journal
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...
OPTIMIZED ROUTING AND DENIAL OF SERVICE FOR ROBUST TRANSMISSION IN WIRELESS N...
IRJET Journal3 views
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in... by IRJET Journal
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...
A Survey on A Secure Anti-Collusion Data Sharing Scheme for Dynamic Groups in...
IRJET Journal17 views
Overlapped clustering approach for maximizing the service reliability of by IAEME Publication
Overlapped clustering approach for maximizing the service reliability ofOverlapped clustering approach for maximizing the service reliability of
Overlapped clustering approach for maximizing the service reliability of
IAEME Publication374 views
IRJET- A Data Stream Mining Technique Dynamically Updating a Model with Dynam... by IRJET Journal
IRJET- A Data Stream Mining Technique Dynamically Updating a Model with Dynam...IRJET- A Data Stream Mining Technique Dynamically Updating a Model with Dynam...
IRJET- A Data Stream Mining Technique Dynamically Updating a Model with Dynam...
IRJET Journal19 views
IRJET- AC Duct Monitoring and Cleaning Vehicle for Train Coaches by IRJET Journal
IRJET- AC Duct Monitoring and Cleaning Vehicle for Train CoachesIRJET- AC Duct Monitoring and Cleaning Vehicle for Train Coaches
IRJET- AC Duct Monitoring and Cleaning Vehicle for Train Coaches
IRJET Journal11 views
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS by ijwscjournal
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONSA COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
ijwscjournal16 views
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS by ijwscjournal
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONSA COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
ijwscjournal9 views
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS by ijwscjournal
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONSA COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
A COMPREHENSIVE SOLUTION TO CLOUD TRAFFIC TRIBULATIONS
ijwscjournal7 views
Optimal Transmission Expansion Planning Using Biogeography... by Leslie Sanchez
Optimal Transmission Expansion Planning Using Biogeography...Optimal Transmission Expansion Planning Using Biogeography...
Optimal Transmission Expansion Planning Using Biogeography...
Leslie Sanchez3 views
IRJET- Cost Effective Scheme for Delay Tolerant Data Transmission by IRJET Journal
IRJET- Cost Effective Scheme for Delay Tolerant Data TransmissionIRJET- Cost Effective Scheme for Delay Tolerant Data Transmission
IRJET- Cost Effective Scheme for Delay Tolerant Data Transmission
IRJET Journal10 views
Accounting Value Effects for Responsible Networking by Giovanni Sileno
Accounting Value Effects for Responsible NetworkingAccounting Value Effects for Responsible Networking
Accounting Value Effects for Responsible Networking
Giovanni Sileno281 views
Traffic Prediction from Street Network images.pptx by chirantanGupta1
Traffic Prediction from  Street Network images.pptxTraffic Prediction from  Street Network images.pptx
Traffic Prediction from Street Network images.pptx
chirantanGupta1489 views
Short Data Rules for Observability.pdf by Dave McAllister
Short Data Rules for Observability.pdfShort Data Rules for Observability.pdf
Short Data Rules for Observability.pdf
Dave McAllister32 views

More from Martin Geddes

Martin Geddes Livestream - March 2018 by
Martin Geddes Livestream - March 2018Martin Geddes Livestream - March 2018
Martin Geddes Livestream - March 2018Martin Geddes
568 views25 slides
Example high-fidelity network measures using ∆Q metrics by
Example high-fidelity network measures using ∆Q metricsExample high-fidelity network measures using ∆Q metrics
Example high-fidelity network measures using ∆Q metricsMartin Geddes
692 views13 slides
Superfast or superfit? The case for UK broadband policy reform by
Superfast or superfit? The case for UK broadband policy reformSuperfast or superfit? The case for UK broadband policy reform
Superfast or superfit? The case for UK broadband policy reformMartin Geddes
491 views22 slides
Broadband service quality - rationing or markets? by
Broadband service quality - rationing or markets?Broadband service quality - rationing or markets?
Broadband service quality - rationing or markets?Martin Geddes
756 views19 slides
The End of Information Technology: Introducing Hypersense & Human Technology by
The End of Information Technology: Introducing Hypersense & Human TechnologyThe End of Information Technology: Introducing Hypersense & Human Technology
The End of Information Technology: Introducing Hypersense & Human TechnologyMartin Geddes
798 views42 slides
The Guardian Avatar by
The Guardian AvatarThe Guardian Avatar
The Guardian AvatarMartin Geddes
6.2K views173 slides

More from Martin Geddes(18)

Martin Geddes Livestream - March 2018 by Martin Geddes
Martin Geddes Livestream - March 2018Martin Geddes Livestream - March 2018
Martin Geddes Livestream - March 2018
Martin Geddes568 views
Example high-fidelity network measures using ∆Q metrics by Martin Geddes
Example high-fidelity network measures using ∆Q metricsExample high-fidelity network measures using ∆Q metrics
Example high-fidelity network measures using ∆Q metrics
Martin Geddes692 views
Superfast or superfit? The case for UK broadband policy reform by Martin Geddes
Superfast or superfit? The case for UK broadband policy reformSuperfast or superfit? The case for UK broadband policy reform
Superfast or superfit? The case for UK broadband policy reform
Martin Geddes491 views
Broadband service quality - rationing or markets? by Martin Geddes
Broadband service quality - rationing or markets?Broadband service quality - rationing or markets?
Broadband service quality - rationing or markets?
Martin Geddes756 views
The End of Information Technology: Introducing Hypersense & Human Technology by Martin Geddes
The End of Information Technology: Introducing Hypersense & Human TechnologyThe End of Information Technology: Introducing Hypersense & Human Technology
The End of Information Technology: Introducing Hypersense & Human Technology
Martin Geddes798 views
Evaluating the internet end-user experience in the Russian Federation by Martin Geddes
Evaluating the internet end-user experience in the Russian FederationEvaluating the internet end-user experience in the Russian Federation
Evaluating the internet end-user experience in the Russian Federation
Martin Geddes605 views
Beasts to superheroes to gods by Martin Geddes
Beasts to superheroes to godsBeasts to superheroes to gods
Beasts to superheroes to gods
Martin Geddes911 views
Beyond 'neutrality' - how to reconnect regulation to reality? by Martin Geddes
Beyond 'neutrality' - how to reconnect regulation to reality?Beyond 'neutrality' - how to reconnect regulation to reality?
Beyond 'neutrality' - how to reconnect regulation to reality?
Martin Geddes2.3K views
The perception gap: the barrier to disruptive innovation in telecoms by Martin Geddes
The perception gap: the barrier to disruptive innovation in telecomsThe perception gap: the barrier to disruptive innovation in telecoms
The perception gap: the barrier to disruptive innovation in telecoms
Martin Geddes2.4K views
Lastingham - Parish mast report by Martin Geddes
Lastingham - Parish mast reportLastingham - Parish mast report
Lastingham - Parish mast report
Martin Geddes669 views
The business world in 2025 by Martin Geddes
The business world in 2025The business world in 2025
The business world in 2025
Martin Geddes2.2K views
A Study of Traffic Management Detection Methods & Tools by Martin Geddes
A Study of Traffic Management Detection Methods & ToolsA Study of Traffic Management Detection Methods & Tools
A Study of Traffic Management Detection Methods & Tools
Martin Geddes1K views
Hypertext to Hypervoice - The next stage in collaboration on the Web by Martin Geddes
Hypertext to Hypervoice - The next stage in collaboration on the WebHypertext to Hypervoice - The next stage in collaboration on the Web
Hypertext to Hypervoice - The next stage in collaboration on the Web
Martin Geddes837 views
Network cost & risk transformation by Martin Geddes
Network cost & risk transformationNetwork cost & risk transformation
Network cost & risk transformation
Martin Geddes747 views
How to get more out of the network? by Martin Geddes
How to get more out of the network?How to get more out of the network?
How to get more out of the network?
Martin Geddes1K views

Recently uploaded

Tunable Laser (1).pptx by
Tunable Laser (1).pptxTunable Laser (1).pptx
Tunable Laser (1).pptxHajira Mahmood
23 views37 slides
ChatGPT and AI for Web Developers by
ChatGPT and AI for Web DevelopersChatGPT and AI for Web Developers
ChatGPT and AI for Web DevelopersMaximiliano Firtman
181 views82 slides
Business Analyst Series 2023 - Week 3 Session 5 by
Business Analyst Series 2023 -  Week 3 Session 5Business Analyst Series 2023 -  Week 3 Session 5
Business Analyst Series 2023 - Week 3 Session 5DianaGray10
209 views20 slides
Attacking IoT Devices from a Web Perspective - Linux Day by
Attacking IoT Devices from a Web Perspective - Linux Day Attacking IoT Devices from a Web Perspective - Linux Day
Attacking IoT Devices from a Web Perspective - Linux Day Simone Onofri
15 views68 slides
STPI OctaNE CoE Brochure.pdf by
STPI OctaNE CoE Brochure.pdfSTPI OctaNE CoE Brochure.pdf
STPI OctaNE CoE Brochure.pdfmadhurjyapb
12 views1 slide
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院 by
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院IttrainingIttraining
34 views8 slides

Recently uploaded(20)

Business Analyst Series 2023 - Week 3 Session 5 by DianaGray10
Business Analyst Series 2023 -  Week 3 Session 5Business Analyst Series 2023 -  Week 3 Session 5
Business Analyst Series 2023 - Week 3 Session 5
DianaGray10209 views
Attacking IoT Devices from a Web Perspective - Linux Day by Simone Onofri
Attacking IoT Devices from a Web Perspective - Linux Day Attacking IoT Devices from a Web Perspective - Linux Day
Attacking IoT Devices from a Web Perspective - Linux Day
Simone Onofri15 views
STPI OctaNE CoE Brochure.pdf by madhurjyapb
STPI OctaNE CoE Brochure.pdfSTPI OctaNE CoE Brochure.pdf
STPI OctaNE CoE Brochure.pdf
madhurjyapb12 views
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院 by IttrainingIttraining
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院
Transcript: The Details of Description Techniques tips and tangents on altern... by BookNet Canada
Transcript: The Details of Description Techniques tips and tangents on altern...Transcript: The Details of Description Techniques tips and tangents on altern...
Transcript: The Details of Description Techniques tips and tangents on altern...
BookNet Canada130 views
Black and White Modern Science Presentation.pptx by maryamkhalid2916
Black and White Modern Science Presentation.pptxBlack and White Modern Science Presentation.pptx
Black and White Modern Science Presentation.pptx
maryamkhalid291614 views
Unit 1_Lecture 2_Physical Design of IoT.pdf by StephenTec
Unit 1_Lecture 2_Physical Design of IoT.pdfUnit 1_Lecture 2_Physical Design of IoT.pdf
Unit 1_Lecture 2_Physical Design of IoT.pdf
StephenTec11 views
HTTP headers that make your website go faster - devs.gent November 2023 by Thijs Feryn
HTTP headers that make your website go faster - devs.gent November 2023HTTP headers that make your website go faster - devs.gent November 2023
HTTP headers that make your website go faster - devs.gent November 2023
Thijs Feryn19 views
handbook for web 3 adoption.pdf by Liveplex
handbook for web 3 adoption.pdfhandbook for web 3 adoption.pdf
handbook for web 3 adoption.pdf
Liveplex19 views
Perth MeetUp November 2023 by Michael Price
Perth MeetUp November 2023 Perth MeetUp November 2023
Perth MeetUp November 2023
Michael Price15 views
6g - REPORT.pdf by Liveplex
6g - REPORT.pdf6g - REPORT.pdf
6g - REPORT.pdf
Liveplex9 views
Data-centric AI and the convergence of data and model engineering: opportunit... by Paolo Missier
Data-centric AI and the convergence of data and model engineering:opportunit...Data-centric AI and the convergence of data and model engineering:opportunit...
Data-centric AI and the convergence of data and model engineering: opportunit...
Paolo Missier34 views
Igniting Next Level Productivity with AI-Infused Data Integration Workflows by Safe Software
Igniting Next Level Productivity with AI-Infused Data Integration Workflows Igniting Next Level Productivity with AI-Infused Data Integration Workflows
Igniting Next Level Productivity with AI-Infused Data Integration Workflows
Safe Software225 views

Advanced network performance measurement techniques

  • 1. Advanced network performance measurement techniques Dr Neil Davies Predictable Network Solutions Ltd Peter Thompson Predictable Network Solutions Ltd Martin Geddes Martin Geddes Consulting Ltd PREDICTABLE NETWORK SOLUTIONS © 2013 All Rights Reserved
  • 2. Dr Neil Davies Co-founder, Predictable Network Solutions Ltd Peter Thompson CTO, Predictable Network Solutions Ltd Martin Geddes Founder, Martin Geddes Consulting Ltd PREDICTABLE NETWORK SOLUTIONS
  • 3. The only ex ante network performance engineering company in the world. Consultancy on the future of telecoms and the Internet. PREDICTABLE NETWORK SOLUTIONS
  • 4. Context for this presentation We are all in the business of “information translocation” The timely movement of information from one computational process to another The value lies in delivering application outcomes That people will pay for You are reading this because you are interested in delivering successful outcomes And understanding the causes of failure, so they can be mitigated You may be working in a culture of deflecting the attribution of blame We’d like to help you turn away from the path to the Dark Side
  • 5. What affects the timeliness? • The timeliness of application outcomes is dependent on the end-to-end loss and delay characteristics of the translocation • We call this end-to-end property ∆Q – ∆Q applies in each direction – not just the round trip – These characteristics need to be suitably bounded • ∆Q depends on the offered load – “Bandwidth” is an aspect of the relationship between offered load and ∆Q This presentation is about measuring ∆Q – and the benefits that approach brings
  • 6. Good measurement is NOT about averages • The average number of legs of a Swedish person is 1.9 – Now find me one! • Measuring average throughput on a 1Gb link over 10 mins is like measuring the traffic on the M5 motorway over two years – No indicator of my likely travel experience • Need to know the instantaneous properties – The ∆Q the “next packet” is going to get • It is all about the probability distribution of quality attenuation – This is what determines timeliness of application outcomes
  • 7. One-point measures • This is the typical information captured by equipment today – Counters (e.g. packets passed, packet sizes, packets dropped) – Sampled over a period • Does not capture ∆Q – Not end-to-end • Multiple one-point measures don’t help • Creates an equipment-centric view – Focuses on the equipment, not the service to the customers – Leads to focus on capacity, and ignores schedulability
  • 8. Multipoint measures • Measure a value between different points – Not just counting things • Same “information translocation” at various points – Measuring the dynamics of the flow • Isolates issues, in both space and time – Excellent diagnostic power • Leads to a focus on schedulability and trading – Which in turn focuses on the outcomes for the customer
  • 9. Different measurement approaches Average Instantaneous Single Point Offered Load and Utilisation (mean values only) Limited predictive power Arrival Patterns Temporal predictive power, localised assurance (compliance with arrival pattern policy) Multiple Point Delay and Loss (mean and variance) Spatial predictive power Temporal and spatial predictive power Assurance of both arrival and service (demand and supply) – represents all that can be known about a system (by observation) PLUS PLUS
  • 10. Interpreting the two-point measure © Predictable Network Solutions 2013 Raw data: There’s no discernable structure here – not possible to work with data like this.
  • 11. Sort by packet size: a clear structure emerges
  • 12. Serialisation (or size-related) delay S Packets with bigger payloads experience more structural delay: it takes longer to turn the packet into a bitstream, and back again into a packet at the next network element . © Predictable Network Solutions 2013
  • 14. Example multi-point measure The bi-directional, end-to-end path of a small cell deployment over commodity infrastructure
  • 15. How to read the information • Different views tell different stories • We’ll see some of those stories in the following slides • The focus on V is because that is where the issues of schedulability manifest themselves
  • 16. Key to following charts Two point measures (by time) GSV view (by packet) V (by time) V (by packet size) V cumulative distribution function (main) V cumulative distribution function (tail)
  • 17. E to A direction (user experience)Return Transit (run dd0a2310-d235-495b-8d2f-a4dc 0 0.05 0.1 0.15 0.2 0 50 100 150 200 250 delay(s) run time (s) Observed Delay against Experiment Run Time E->A 0 0.05 0.1 0.15 0.2 0 delay(s) 0.02 0.04 0.06 0.08 0.1 0.12 0.14 delay(s) Observed Delay Variability (V) against Experiment Run Time E->A 0.02 0.04 0.06 0.08 0.1 0.12 0.14 delay(s) Note the delay spike during the test run @ approx 60 seconds in How can be begin to analyse this performance issue?
  • 18. E -> A (by packet) This ‘spike’ doesn’t appear to be related to a particular packet size (note ‘striations’ in the S value is an artefact of 3GPP scheduling)
  • 19. E -> A (Dynamic response) Removing G and S influences clearly highlights the magnitude of the contention issue
  • 20. Spatial Isolation Same magnitude issue between D to B But not between D and C
  • 21. Spatial Isolation (2) It is occurring between C and B NOTE: this is the effect that we are measuring – NOT the cause (which in this case was not the access network but elsewhere) Armed with this information, we can begin to analyse root causes (e.g. what is over-driving this link?)
  • 22. ∆Q for ADSL line © Predictable Network Solutions 2013 Compare and contrast: baseline data for an ADSL line
  • 23. ∆Q for Femto (over ADSL) © Predictable Network Solutions 2013 Now run a femtocell over that same line: much worse performance
  • 24. Summary • Multipoint distribution based measurement gives access to all the information available through observation – “observation” is key – independent of equipment – Captures the influence of technology etc • G,S & V gives you a way of extracting both temporal and spatial details • Becomes extremely powerful when combined with analysis – E.g. you have a model of what V should be, or what G and S should be given the network layout
  • 25. Upcoming workshops: Sustainable Public Service Networks London, 19th September 2013 Fundamentals of Network Performance London, 20th September 2013 www.sustainablebroadband.com PREDICTABLE NETWORK SOLUTIONS
  • 26. Neil Davies Neil.Davies@pnsol.com Peter Thompson Peter.Thompson@pnsol.com Martin Geddes mail@martingeddes.com PREDICTABLE NETWORK SOLUTIONS

Editor's Notes

  1. There is a branch of mathematics called Large Deviation Theory that does have something to say about the predictive power of averages. And what it says is not very comforting – in the sense what it means in terms as a predictor of underlying hazards and risks – i.e. it is pretty bad. Capturing distribution gives the ability to assure that arrival patterns are within specification (see QTA’s later)Multipoint measurement gives some level of spatial identification – but suffers from the same issues as A, in that it remains a bad predictor of the hazards and risks.This is the measurement nirvana – it turns out that multipoint instantaneous observation makes available all the information that is possible by observation. In a well designed system this, by the principle of observational bisimularity, is the ultimate evidence of correct operation – including its performance aspects.Although there are people who have prided themselves on capturing (average) data over smaller (and smaller) timescales – the real issue is the number of events that occur in those timeslots. This is where the M5 (major holiday highway in UK) analogy comes in – the number of possible “events” 10 minutes on a 1Gb Ethernet (i.e. packets) is broadly equivalent to the number of events (cars on a three lane highway) in two year. For “averaging” to make sense you would need to be generating them over 20ms to 250ms intervals – no one can afford that.
  2. Note that S is not a number, it is a function from packet size to delay. S is not necessarily a simple line, it may have a more complex structure depending on media quantisation (eg ATM cells, WiFi) and bearer allocation choices (eg 3GPP).