Top Banner
Tracking Bulgarian Internet evolution from 2001 till now Nikolay Milovanov Vasil Yordanov https://github.com/iTrans formers
27
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: networkEvolution

Tracking Bulgarian Internet evolution from

2001 till now Nikolay Milovanov

Vasil Yordanovhttps://github.com/iTransformers

Page 2: networkEvolution

Network evolution in nutshell

• Network evolution could be expressed as:– Topology changes:

• Appearance, disappearance of nodes or links• Changes in network diameter• Changes in node ranks • Changes in clustarization and giant components

– Changes in node/edge properties• Appearance of new metadata• Disappearance of such• Changes in node/edge metadata values

– All the above reflects changes in the law as per which network evolves :• An example of such law is Barbatsi and Albert law – e.g human

networks evolve as per power law

Page 3: networkEvolution

Examples for evolving networks

• Internet• WWW• Social• Bio• Communication• Networks of SA components and connectors

• Knowing the law will allow us to predict how does the network will evolve

Page 4: networkEvolution

So a question emerge: How can we track how does the network evolves?

– Globul OMC used a notepad to write down each and every event that happens in the network

– Traffic/Performance monitoring systems might give us such information but in really it is quite difficult to abstract it (too much details)

– Well we don’t care so much of the tiny networking details we care for evolution on a architecture level!

Page 5: networkEvolution

Views and perspectives• Many stakeholders result in many perspectives and thus many

views of the network are required!

The cardiologist needs a cardiogram

.. But the orthopedist needs an X-ray scan

Stakeholders needs views of the network that will allow them to reason about it!

Page 6: networkEvolution

In more common language

• Core network engineers needs a view of the core network, they don’t care much about the access…

• BGP peering needs a view of the BGP they don’t care much neither about the core nor about the access!

• Support guys from Plovdiv field maintenance team needs a view of the network from their region, they don’t care about Sofia network

Page 7: networkEvolution

Initial questions

• How does the number of Bulgarian Internet Autonomous systems evolve through the years (totals and node births/deaths)

• How did the number of Bulgarian Internet BGP peering to the rest of the world changed over the years?

• What is the number of total advertised addresses?

• All that has to be done for IPv4 & IPv6

Page 8: networkEvolution

BG INTERNET IPV4

Page 9: networkEvolution

Bulgarian IPv4 autonomous systems (totals per year)

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

100

200

300

400

500

600

Page 10: networkEvolution

International BGP Peering BG to *

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

20

40

60

80

100

120

Page 11: networkEvolution

Bulgarian IPv4 address space

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

1000000

2000000

3000000

4000000

5000000

6000000

7000000

Page 12: networkEvolution

IPv4 AS birth /dead Achieved by diffing per year

2001-2002

2002-2003

2003-2004

2004-2005

2005-2006

2006-2007

2007-2008

2008-2009

2009-2010

2010-2011

2011-2012

2012-2013

2013-2014

0

10

20

30

40

50

60

Node birthsNode deads

Page 13: networkEvolution

BG INTERNET IPV6

Page 14: networkEvolution

BG IPv6 ASNs

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

10

20

30

40

50

60

Page 15: networkEvolution

International IPv6 peering

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

5

10

15

20

25

30

35

Page 16: networkEvolution

Advertised IPv6 Address Space

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 M

50000 M

100000 M

150000 M

200000 M

250000 M

300000 M

Page 17: networkEvolution

BG IPv6 AS birth /dead Achieved by diffing per year

2001-2002

2002-2003

2003-2004

2004-2005

2005-2006

2006-2007

2007-2008

2008-2009

2009-2010

2010-2011

2011-2012

2012-2013

2013-2014

0

2

4

6

8

10

12

Page 18: networkEvolution

HOW DID WE DO IT?

Page 19: networkEvolution

The approach -1

• We have generated Global IPv4/IPv6 Internet maps from MRT (RFC 6396) files.

• The raw data from the repository of the route views project ftp://archive.routeviews.org/

• Each MRT file will contain full Internet routing table. • We will extract BGP AS PATH attributes and will generate a map based on

Internet Autonomous System numbers. • As part of the map generation process each BGP AS has been enriched by

whois info from http://www.potaroo.net/bgp/iana/asn.txt)• The maps itself are generate d with javaMrt2Graphml https://github.com/iTransformers/javaMrt2Graphml• We will generate one map per year (lazy and statistically incorrect) • Each map will contain the “whole” Internet

Page 20: networkEvolution

The approach 2

• Each Internet map sub-graph will be filtered by netTransformer to a BG Internet sub-graph.

• By BG Internet we mean Bulgarian ASNs and their peering to the rest of the world (e.g Bulgaria connectivity to abroad).

• With netTetTransformer we will generate graph diffs on yearly by year basis (e. 2001-2002, 2002 – 2003 and so on)

• Finally will calculate node rankings and will generate diff reports

• All that was used for the presented statistics generation

Page 21: networkEvolution

netTransformer

• Software for tracking and pushing the network evolution process

• Network transformation – a process in which we push the network evolution in direction that suits our stakeholder interest

• It is able to discover networks, to fill in different network models, to diff between any two discovered network states

• It is able also to “talk” with network devices and to executes the steps of the strategies for network from state to state transformation.

Page 22: networkEvolution

netTransformer

• Written in java, xslt, glue with maven, groovy, Spring, JUNG and a lot of swing ;)

• https://github.com/iTransformers/netTransformer• GPL!• Architecture is based on the following key

Interfaces/components– Discoverers (anybody can implement his own, we have a couple raging

from SNMP, Openflow floodlight, BGP peering and some others). – Node or Network Listeners– Viewer, resourceFactories, activationFactories – postDiscoverers (asses networks after the discovery has finished!)

Page 23: networkEvolution

netTransformer Look and feels

Page 24: networkEvolution

Internet BGP AS Node MetadataAutonomous system ID

IPv6 advertised prefix count

AS name & description (whois)

Advertised IPv4/IPv6 prefixes (comma separated)

IPv4/IPv6 address count

Country of affiliation

IPv4 advertised prefix count

Page 25: networkEvolution

Graph Diffing

Page 26: networkEvolution

A few words about OpenSource and our motivation

• The surprising truth about what motivates ushttps://www.youtube.com/watch?v=u6XAPnuFjJc

Page 27: networkEvolution

QUESTIONS!

Nikolay MilovanovVasil Yordanovhttps://github.com/iTransformers/

Tracking Bulgarian Internet evolution from 2001 till now