36
NETWORK MEASUREMENT VIRTUAL OBSERVATORY István Csabai Eötvös University Budapest

István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Embed Size (px)

Citation preview

Page 1: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

NETWORK MEASUREMENT VIRTUAL OBSERVATORY

István Csabai Eötvös University Budapest

Page 2: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

Science: early times

Page 3: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

test

predictions

Science: past

Page 4: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

virtual realitypredictions

test

Science: present

Page 5: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

virtual realitypredictions

test

Other disciplines are similar: whole genomes, satellite maps, sensor networks, etc.

Page 6: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

To understand the complex reality, we need complex models

To verify complex models we need a lot of data (large statistical sample)

Page 7: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Virtual Observatories: few key elements

Modern web and database technologies to standardize and share data

Advanced indexing techniques for multi-terabyte, multidimensional archives

Distributed computing for analysis and modeling – move computation to data

Visualization Same challenges for different

disciplines: we can share the tools and solutions

Page 8: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Internet

Although man-made but there is no “blueprint”

“Astronomical” number of components, complex non-linear interactions

We need similar methods as in natural sciences to understand its behavior

We need to observe / experiment model

Future internet: self-aware, self-managing, self-healing …

Page 9: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Measurement/experiment facilities Planetlab / Onelab

A network of open computers distributed across the world and available for the development of new network services.

Realistic platform available for trial deployment and experimentation, with services such as distributed storage, network mapping, peer-to-peer systems, distributed hash tables and query processing.

OneLab2 EU Integrated Project 2008-2010 ,29 participants Europe wide + 1 from Japan, budget 7.5 M€

Page 10: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Measurement/experiment facilities Etomic

An open measurement infrastructure in Europe to carry out high temporal resolution (~10 nano second), globally synchronized, active measurements. Provides high resolution, spatially extended dynamic picture of fast changes in the network traffic.

Page 11: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

ETOMIC hardware Server PC architecture, Linux Endace DAG 3.6 GE card

or ARGOS FPGA measurement card with packet sending capability (packet offset

~60 ns) GPS antenna for

global time synchronization

New low cost version based on Blackfin microcontroller dedicated IP packet time stamping module

(< μsec) Can be called as SOAP web services / from

stored procedures low cost (300 €)

Page 12: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Web interface: experiment bundle

Page 13: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Scheduling

Experiments

Time slot reservation in a calendar system for the experiments

Page 14: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

User experiments

Page 15: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Science archive – based on CasJobs

CasJobs: database + batch + web services(Developed for SDSS; Alex Szalay , William O’Mullane, Nolan Li, María Nieto-Santisteban, Ani Thakar, Jim Gray)

Page 16: István Csabai Eötvös University Budapest. observation theoryreality Science: early times
Page 17: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

most network measurement projects:

• use a single dedicated infrastructure

• scan only narrow sub-segments

• analyze a limited set of networkcharacteristics

• centralized and separated fromeach other

key idea: try to interconnect separate• measurement efforts

• large-scale behavior

• long-term evolution

Network Measurement Virtual Observatory

Page 18: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Network Measurement Virtual Observatory The challenge: seamless integration of multiple tools

and data sources We need easy to use standards for data access/exchange Searchable metadata, organized access Working prototypes, use cases

Unified interface: standardized data model (NetXML) wrapper software @ each platform (to convert data into

NetXML) standardized way of reaching services (Web services)

Mediator: metadata registry service registry (e.g. time slice service) joint queries: query broker Many ideas, concepts from

astronomical VO

Page 19: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

database

Processing functionsStored procedures

Search toolsVisualization tools

Web service

interface

graphical user interface

traceroute archive

SQL

data subset

XML, SOAP

WSDL, processed data

database

Processing functionsStored procedures

Search toolsVisualization tools

Web service

interface

graphical user interface

delay archive

SQL

data subset

XML, SOAP

WSDL, processed data

database

Processing functionsStored procedures

Search toolsVisualization tools

Web service

interface

graphical user interface

BGP archive

SQL

data subset

XML, SOAP

WSDL, processed data

Mediator/query broker

GUI

WS

Page 20: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Use cases: what do we measure

one way delay (60 ns resolution)

tracking topology changes available bandwidth meter transport protocol testing queuing delay tomography geolocalization …

Page 21: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Geolocalization

A

B

?

Page 22: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Geolocalization Useful for: content distribution, user identification,

network diagnostics, language selection,… WhoIS and DNS databases are not reliable Method: probabilistic triangulation from landmark

nodes More landmarks – more precise localization – more

computation

Page 23: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Hierarchical Triangular MeshHierarchical subdivision of spherical triangles, represented as quad tree.

Developed for indexing the sky (SDSS; A. Szalay, G. Fekete)

Standalone library + SQL Server integrationQuick routines for intersecting spherical polygons: speedup geolocalization calculations

Page 24: István Csabai Eötvös University Budapest. observation theoryreality Science: early times
Page 25: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

virtual realitypredictions

test

Conclusion

Page 26: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

virtual realitypredictions

test

Conclusion

Page 27: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

observationtheory reality

models

experiment

instruments

virtual realitypredictions

test

Conclusion

Page 28: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

P. Mátray, P. Hága, S. Laki, J. Stéger B. Hullár, G. Vattay (Eötvös University)

collaborations

Universidad Autonoma de Madrid Universidad Publica de Navarra Ericsson Research Tel Aviv University

This work was partially supported by the National Office for Research and Technology (NAP 2005/ KCKHA005) and the EU ICT Moment (#215225) and OneLab2 Integrated Project

(#224263).

Collaborators, acknowledgements

Page 29: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Sok boldog születésnapot, Sanyi!

Page 30: István Csabai Eötvös University Budapest. observation theoryreality Science: early times
Page 31: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Sándor Laki (C) IP Geolocation 31

Plans

Probabilistic geolocation Using the distribution of the velocity

of signal propagation Pre-clustering delay measurements

Page 32: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

New kind of science: mega-surveys New technology has allowed new kind of surveys

All-sky surveys GALEX ,SDSS, IRAS,WMAP, NVSS, …

Multi-wavelength Radio, infrared, optical, UV, X-ray, Gamma-ray Gravitational waves, neutrinos & cosmic particles

Time domain Pan-STARRS, LSST, GAIA, …

Collecting data in (public) digital archives Instead of targeting special objects: observe

everything New challenge: store, organize, analyze huge amount of

data Virtual Observatories / E-science

Page 33: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

main goals:• create large archives to:

share available measurement data & analysis results provide easy-to-use “online” network data analysis tools

• integrate existing measurement and monitoring infrastructures toward a common and open platform:

standardisation of network measurement data design and implement a system to share network data

Page 34: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

store & share raw data• joint analysis of different types of measurement data• reanalysis (with new statistical methods)• reference data (historical comparison)

share analysis tools• server side processing simplifies client applicatons• no need to transfer bulk data packages: online processing

exchange data in a standardized way• existing efforts (OGF, IETF, KML)

• Network Measurement Virtual Observatory (nmVO)

Database

Web servicesnetXML

Page 35: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Science archive – based on CasJobs

database Web service

interface Stored procedure

Page 36: István Csabai Eötvös University Budapest. observation theoryreality Science: early times

Open repository