Download 990mplanedagstuhl

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Neuroinformatics wikipedia , lookup

Data analysis wikipedia , lookup

Psychometrics wikipedia , lookup

Transcript
mPlane – Building an Intelligent
Measurement Plane for the Internet
Alessandro D’Alconzo
FTW Forschungszentrum Telekommunikation Wien, AT
[email protected]
IFIP TC6 2014/2 Strategic Review Meeting, Dagstuhl, Germany
November 12-14, 2014
mPlane project quick facts

mPlane is an FP7 Integrated Project

Project acronym: mPlane
Project full title: “mPlane – an Intelligent Measurement
Plane for Future Network and Application Management”
Grant agreement no: 318627
Staring Date: November 1st 2012
Total Cost: 11,274,908.00 €
Duration: 3 years
Partners: 16
Coordinator: Prof. Marco Mellia – Politecnico di Torino - IT







2
Who we are




3 operators
6 research centers
5 universities
2 small enterprises
Coordinator
Marco Mellia
POLITO
Ernst Biersack
Eurecom
Andrea Fregosi
Fastweb
Guy Leduc
Univ. Liege
Tech. Coordinator
Dina Papagiannaki
Telefonica
Saverio Nicolini
NEC
Brian Trammell
ETH
Tivadar Szemethy
NetVisor
Dario Rossi
ENST
Fabrizio Invernizzi
Telecom Italia
Pedro Casas
FTW
Pietro Michiardi
Eurecom
3
The mPlane community

Many Academic and research institutions come from the
Traffic Monitoring Analysis (TMA) COST Action:


PoliTo, ETH, ENST, Eurecom, UniLiege, FTW, NEC
Some others were part of previous EU projects:




PRISM: FTW, ETH
NapaWine: PoliTo, ENST, NetVisor, NEC
DEMONS: FTW, ETHZ, NEC, TID
…
4
Which problem(s) mPlane
aims at solving
The nowadays Internet
“The Internet is the first thing that humanity has
built that humanity doesn't understand, the
largest experiment in anarchy that we have ever
had.”
Eric Schmidt – ex Google Exec. Chairman
6
A complicated technology…
 Internet: different technologies are combined to offer a plethora
of services
 We sorely miss the technology to understand what is
happening in the network and thus to optimize its performance
and utilization
 Specially when something goes wrong!
7
A complicated technology…
…that no one controls and understands




Why
is not working?
Which is the best ISP in my area?
Where is
traffic coming from?
How to optimize my
network for
facebook?
8
Why we are here: mPlane
motivation

The Internet is a global interconnection of networks



In case of “failure”, who can tell what’s going wrong?



Each ISP may have a figure of what happens inside its network
But what if the failure depends on other ISPs? Or on the content
provider? Or on the CDN? Or on user equipment?
Today, the web is a tangle



No single organization operates, administers or governs it
It is robust thanks to its diversity, but it is vulnerable and fragile
Nobody really understands what happens today in the Internet
How to predict what will happen tomorrow?
We need an intelligent system that collects, analyzes, provides
visibility to support better management: an oracle that provides
answers
9
How can mPlane solve the
problem(s)?
The mPlane vision

Goal: design and demonstration of a measurement
plane for the Internet




A distributed infrastructure for network measurement
… which perform passive and active measurements,
continuously or on-demand, at a wide variety of scales
… with built-in support for iterative measurement and
automated iteration.
mPlane is about



large scale network measurements,
and intelligent big-data analysis for troubleshooting
support
embedding measurement into the Internet as an
additional capability
11
mPlane in a slide
Build a distributed, open, standard measurement
infrastructure for the Internet

Probes (WP2) – get the data



Build on existing tools/methodologies
Offer a flexible, programmable, open platform to run and collect
Pietro Michiardi
passive, active, hybrid measurement
Repositories (WP3) – store and preprocess the data




Tivadar Szemethy
Collect measurement in a standard way
Pre-process large amounts of data in efficient ways
Grant access to interested parties (ISP, content providers, end-users,
regulation agencies, etc.) subject to authorization rules
Pedro Casas
Intelligent reasoner (WP4) – dig into the data


Mine automatically the data and extract useful information
Help in drilling down to the root cause of a problem
12
mPlane WPs’organization
WP8 - Project Management
WP7 - Dissemination, Exploitation and Standardization
WP4 - mPlane Supervisor: Iterative
and Adaptive Analysis (supervision layer)
WP3 - Large-scale Data Analysis
(Repository and Analysis Layer)
WP2 – Programmable Probes
(Measurement Layer)
WP1
Use Cases,
Requirements
and Architecture
Ruediger
Martin
WP5
Integration,
Deployment,
Data
Collection,
Evaluation
WP6
Demonstration
13
mPlane in a picture
Coordination
Intelligent
Reasoner
WP4
Module N
Module 2
Module 1
Analysis Modules
Repository and Analysis Layer
Data collection
& processing
Supervisor
legacyDB 1
legacyDB 2
legacyDB N
mPlane
Repository
WP3
DBStream
Blockmon
Measurement Layer
WP2
mInterface
mInterface
Raw
data
mInterface
mProbe 1
mProbe 2
mProbe N
mInterface
mInterface
legacyProbe 1 legacyProbe 2
mInterface
e
legacyProbe N
14
Some of the mPlane Use Cases

Cloud services troubleshooting

Mobile network performance issues troubleshooting

Web browsing QoE troubleshooting

Anomaly detection and diagnosis in large scale netoworks

Multimedia content delivery troubleshooting

Content and service popularity estimation & caching for network
optimization

SLA verification and certification
15
Who benefits from mPlane?

mPlane benefits everyone:

ISPs get a fine-grained picture of the network status, empowering
effective management, operation, and troubleshooting.

Content and Application providers gain powerful tools for
handling performance issues of their delivery systems and
applications.

Regulators and end-users can verify adherence to SLAs, even
when these involve many parties.

Customers of all kinds can objectively compare network
performance, improving competition in the market.

The Research Community gets a system to accelerate the pace
of research driven by Internet measurements
16
Collaborating Institutions

mPlane promotes collaboration with external partners




Official list of collaborating institutions







To strengthen the standardization effort in mPlane
To allow larger deployment of the mPlane system
To enable external partners to get in touch with mPlane technology
Univ. Federal de Juiz de Fora (Prof. Borges Vieira) , BR
Univ. Federal de Minas Gerais (Prof.s Marques de Almeida and Couto da Silva), BR
Univ of Auckland (Nevil Brownlee), NZ
Univ. of Wurzburg (Prof. Tran-Gia, Tobias Hossfeld), DE
Orange Polska S.A (Dr Zbigniew Kopertowski), PL
University of Twente (Aiko Pras, Anna Sperotto), NL
Active Collaborations




University of Waterloo (Prof. Lukasz Golab), CND
Cooperative Association for Internet Data Analysis (Dr Alberto Dainotti), US
Northwestern University (Prof. Fabian Bustamante), US
Boston University (Prof. Mark Crovella), US
17
Dissemination activities (1)

Schools

Summer school on "Big-Data Algorithms and Applications to Traffic
Measurements” in collaboration with BigFoot project (bigfootproject.eu)



4th TMA PhD School, in collaboration with ACM SIGCOMM, March 15-16th
2014, London



On mPlane architecture and coding of mPlane-compatible monitoring
probes during the METRICS bootcamp

Aug. 25 – Sept. 5, 2014, Université Catholique de Louvain, Belgium
Seminar

Contribution to the organisation of the Dagstuhl Seminar on "Global
Measurement Framework".


Organizers: Dario Rossi – ENST
Speakers from ENST, PoliTo, Uni Liege, Inria, Uni Pierre and Marie Curie
Workshop


Organizers: Marco Mellia – PoliTo, Dario Rossi - ENST
Speakers: Pietro Michiardi, Antonio Barbuzzi – Eurecom
Organizers: Philip Eardley (BT Research R&D, GB), Marco Mellia (PoliTo, IT), Jörg Ott
(Aalto University, FI), Jürgen Schönwälder (Jacobs University Bremen, DE), Henning
Schulzrinne (Columbia University, US)
20 invited talks and project presentations
18
Dissemination activities (2)

Conferences




Gold sponsor of ACM IMC 2013, organized by TID in Barcelona
TMA 2013, in conjunction with IEEE Infocom in Turin, and chaired by
mPlane partners
Several mPlane partners involved in the organization of ACM IMC,
CoNEXT, IEEE Infocom, TMA, IEEE TRAC
Standardization

IETF: ETH active in several working groups and meetings




Brian Trammel from ETH is the chair of the IETF IP Performance Metrics (IPPM)
Contributes to Large-Scale Measurement of Broadband Performance (LMAP),
Congestion Exposure (ConEx), and TCP Maintenance and Minor Extensions
(tcpm) WGs
3 RFCs have being published during the project
5 RFCs drafts are in progress
 Full list at: http://www.ict-mplane.eu/public/standardization


IRTF: A-LBELL actively contributes to the Network Management
Research Group (NMRG)
Brian Trammel appointed as Internet Architecture Board (IAB) member
on February 2014
19
Other ongoing efforts for measurement frameworks

FP7 European projects



Integrated Project (IP)
3 years  2 left, 16 partners, 11.2 Meuros
… is like a “mPlane use case”
“From global measurements to local management”

Specific Targeted Research Projects (STReP)

3 years  2 left, 10 partners, 3.8 Meuros

Build
probes
Strong similarities
for a measure framework out of

the architecture core

IETF, Large-Scale Measurement of Broadband Performance (LMAP)



Standardization effort on how to do broadband measurements

Defining the components, protocols, rules, etc.
It does not specifically target adding “a brain” to the system
Brian Trammell
ETH
Initial discussion with institutions such as RIPE and CAIDA
20
Where we are


We are at the end of 2nd year 
We have lot of “tools”



Active, passive, mobile, browsers, edge, core, …
We have lot of research papers in the most prominent
conferences, including ACM IMC, IEEE Infocom, PAM,
IFIP TMA, IEEE TRAC
st
 47 items during 1 year
 25 items during 9 months of 2nd year
 2 Awards (ANRP 2013, ITC25)
…
https://www.ict-mplane.eu
21
Conclusions

mPlane aims at simplifying network monitoring practices
 Reasoning and Analysis focused on iterative
measurements


Open source release of software


Tstat, Blockmon, QoF, Tracebox, etc.
Maximum reuse of existing measurement tools


Troubleshooting support
First software libraries available at the mPlane website
Open for collaborations

Collaboration Institutions (CI)


Mlab, Orange Lab Poland, Endace, etc.
Other (less formal) ways are welcomed as well 
22
Thank you for Your
attention!
Alessandro D’Alconzo – FTW
[email protected]