On multi-exit routings and AS relationships Riad Mazloum , - - PowerPoint PPT Presentation

▶
on multi exit routings and as relationships
SMART_READER_LITE
LIVE PREVIEW

On multi-exit routings and AS relationships Riad Mazloum , - - PowerPoint PPT Presentation

Nov 12, 2022 107 likes •448 views

On multi-exit routings and AS relationships Riad Mazloum , Marc-Olivier Buob 1 , Jordan Aug e 1 , Bruno Baynat 1 , Timur Friedman 1 and Dario Rossi 2 1 UPMC, France first.last@lip6.fr 2 Telecom ParisTech, France dario.rossi@enst.fr February

slide-1
SLIDE 1

On multi-exit routings and AS relationships

Riad Mazloum, Marc-Olivier Buob1, Jordan Aug´ e1, Bruno Baynat1, Timur Friedman1 and Dario Rossi2

1UPMC, France

first.last@lip6.fr

2Telecom ParisTech, France

dario.rossi@enst.fr February 06th, 2013 –ISMA 2013 AIMS 5

On multi-exit routings and AS relationships 1

slide-2
SLIDE 2

Internet routing example

On multi-exit routings and AS relationships 2

slide-3
SLIDE 3

Internet routing example

On multi-exit routings and AS relationships 3

slide-4
SLIDE 4

Surprising observations

On multi-exit routings and AS relationships 4

slide-5
SLIDE 5

BGP route with higher LP

On multi-exit routings and AS relationships 5

slide-6
SLIDE 6

A route has a higher LP

On multi-exit routings and AS relationships 6

slide-7
SLIDE 7

Single next-hop AS

On multi-exit routings and AS relationships 7

slide-8
SLIDE 8

Multi next-hop ASes

On multi-exit routings and AS relationships 8

slide-9
SLIDE 9

Equal LP for all next-hop ASes

On multi-exit routings and AS relationships 9

slide-10
SLIDE 10

Observed ME = ⇒ equal LP for next-hop ASes

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

On multi-exit routings and AS relationships 10

slide-11
SLIDE 11

Implementation of AS economical policies

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider

On multi-exit routings and AS relationships 11

slide-12
SLIDE 12

Implementation of AS economical policies

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider = ⇒ LP(client) > LP(peer) > LP(provider)

On multi-exit routings and AS relationships 11

slide-13
SLIDE 13

(ME) + (POLICY)

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider = ⇒ LP(client) > LP(peer) > LP(provider)

(ME) + (POLICY)

ME(AS1, p) = {AS2, AS3} = ⇒ same type of relationship between AS1 and AS2, AS3

On multi-exit routings and AS relationships 12

slide-14
SLIDE 14

Does it work?

(ME) + (POLICY)

ME(AS1, p) = {AS2, AS3} = ⇒ same type of relationship between AS1 and AS2, AS3

On multi-exit routings and AS relationships 13

slide-15
SLIDE 15

Does it work?

(ME) + (POLICY)

ME(AS1, p) = {AS2, AS3} = ⇒ same type of relationship between AS1 and AS2, AS3

Check with CAIDA’s inference dataset

  • Get types of relations of cases such between AS1 and

AS2, AS3 from CAIDA’s inference dataset

  • Check whether all of the next-hop ASes have the same

relationship

On multi-exit routings and AS relationships 13

slide-16
SLIDE 16

Does it work?

(ME) + (POLICY)

ME(AS1, p) = {AS2, AS3} = ⇒ same type of relationship between AS1 and AS2, AS3

Check with CAIDA’s inference dataset

  • Get types of relations of cases such between AS1 and

AS2, AS3 from CAIDA’s inference dataset

  • Check whether all of the next-hop ASes have the same

relationship

  • About 70% matching 30% mismatching

On multi-exit routings and AS relationships 13

slide-17
SLIDE 17

Multi-exit occurrences and relationship matching

20 40 60 80 100 2 3 4 5 6 7 8 9 10 11 12 13 Percentage of matching # next-hop ASes Percentage of mathcnig between multi-exit nex-hop relationships and CAIDA inference 1 10 100 1000 10000 100000 1e+06 2 3 4 5 6 7 8 9 10 11 12 13 # occurences of ME # next-hop ASes Occurences of multi-exit routing Total number of multi-exits On multi-exit routings and AS relationships 14

slide-18
SLIDE 18

Data

Multi-exit discovery:

  • BGP: BGPmon, Colorado State University project1

1http://bgpmon.netsec.colostate.edu/ 2http://www.top-hat.info/ 3http://www.team-cymru.org/Services/ip-to-asn.html 4http://www.caida.org/data/active/as-relationships/ On multi-exit routings and AS relationships 15

slide-19
SLIDE 19

Data

Multi-exit discovery:

  • BGP: BGPmon, Colorado State University project1
  • traceroute: TDMI/TopHat, UPMC project2
  • IP/AS aliasing: Team Cymru IP to AS mapping service3

1http://bgpmon.netsec.colostate.edu/ 2http://www.top-hat.info/ 3http://www.team-cymru.org/Services/ip-to-asn.html 4http://www.caida.org/data/active/as-relationships/ On multi-exit routings and AS relationships 15

slide-20
SLIDE 20

Data

Multi-exit discovery:

  • BGP: BGPmon, Colorado State University project1
  • traceroute: TDMI/TopHat, UPMC project2
  • IP/AS aliasing: Team Cymru IP to AS mapping service3

AS relationships:

  • CAIDA AS relationship inference database4

Our data is available on request.

1http://bgpmon.netsec.colostate.edu/ 2http://www.top-hat.info/ 3http://www.team-cymru.org/Services/ip-to-asn.html 4http://www.caida.org/data/active/as-relationships/ On multi-exit routings and AS relationships 15

slide-21
SLIDE 21

Discussion

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider = ⇒ LP(client) > LP(peer) > LP(provider)

Check with CAIDA’s inference dataset

  • Get types of relations of cases such between AS1 and

AS2, AS3 from CAIDA’s inference dataset

  • Check whether all of the next-hop ASes have the same

relationship

On multi-exit routings and AS relationships 16

slide-22
SLIDE 22

Discussion

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider = ⇒ LP(client) > LP(peer) > LP(provider)

Check with CAIDA’s inference dataset

  • Get types of relations of cases such between AS1 and

AS2, AS3 from CAIDA’s inference dataset

  • Check whether all of the next-hop ASes have the same

relationship

On multi-exit routings and AS relationships 17

slide-23
SLIDE 23

Discussion

Observed ME = ⇒ equal LP (ME)

ME(AS1, p) = {AS2, AS3} = ⇒ LPAS1(AS2) = LPAS1(AS3)

Implementation of AS economical policies (POLICY)

client > peer > provider = ⇒ LP(client) > LP(peer) > LP(provider)

Check with CAIDA’s inference dataset

  • Get types of relations of cases such between AS1 and

AS2, AS3 from CAIDA’s inference dataset

  • Check whether all of the next-hop ASes have the same

relationship

On multi-exit routings and AS relationships 18

slide-24
SLIDE 24

Have another ideas?

  • Feedback about the problem and the analysis process
  • Get confirmation about the results (we don’t have a ground

truth of AS relationships)

  • Possible collaborations

On multi-exit routings and AS relationships 19

slide-25
SLIDE 25

Reserved slides...

On multi-exit routings and AS relationships 20

slide-26
SLIDE 26

BGP and IP overlap

On multi-exit routings and AS relationships 21

slide-27
SLIDE 27

Multi-exit routing, BGP and IP results

On multi-exit routings and AS relationships 22

slide-28
SLIDE 28

TopHat interconnection

On multi-exit routings and AS relationships 23

slide-29
SLIDE 29

On multi-exit routings and AS relationships 24

slide-30
SLIDE 30

On multi-exit routings and AS relationships 25

slide-31
SLIDE 31

On multi-exit routings and AS relationships 26

slide-32
SLIDE 32

On multi-exit routings and AS relationships 27