From Trustee Georgia to Royal Georgia Georgia Studies Notes.
Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia...
-
Upload
douglas-bryan -
Category
Documents
-
view
217 -
download
0
Transcript of Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia...
![Page 1: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/1.jpg)
1
Analysis of peering strategy adoption by transit providers in the Internet
Aemen Lodhi (Georgia Tech)
Amogh Dhamdhere (CAIDA)
Constantine Dovrolis (Georgia Tech)
2nd Workshop on Internet Economics (WIE’11)
![Page 2: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/2.jpg)
2
Outline
• Motivation• Objective• The model: GENESIS• Results– Adoption of peering strategies– Impact on economic fitness–Who loses? Who gains?
• Alternatives• Conclusion
![Page 3: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/3.jpg)
3
Motivation• Peering strategies of ASes in the Internet (source: PeeringDB www.peeringdb.com)
![Page 4: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/4.jpg)
4
Motivation• Peering policies for networks classified by traffic
ratio (source: PeeringDB www.peeringdb.com)
![Page 5: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/5.jpg)
5
Objective
• Why do transit providers peer Openly?
• Impact on their economic fitness?• Which transit providers lose/gain? • Are their any alternative peering
strategies?
Enterprise
customer
Transit Provide
r
Transit Provide
r
Internet
Enterprise
customer
Content Provider
Content Provider
![Page 6: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/6.jpg)
Approach
• Agent based computational modeling• Scenarios
Conservative• Selective • Restrictive
Non-conservative
• Selective• Restrictive• Open
vs.
![Page 7: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/7.jpg)
7
The model: GENESIS*
• Agent based interdomain network formation model
• Incorporates– Co-location constraints in provider/peer
selection– Traffic matrix– Public & Private peering– Set of peering strategies– Peering costs, Transit costs, Transit
revenue*GENESIS: An agent-based model of interdomain network formation, traffic flow and economics. To appear in InfoCom'12
![Page 8: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/8.jpg)
The model: GENESIS*
Fitness = Transit Revenue – Transit Cost – Peering cost
• Objective: Maximize economic fitness• Optimize connectivity through peer
and transit provider selection• Choose the peering strategy that
maximizes fitness
![Page 9: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/9.jpg)
9
Peering strategies
• Restrictive: Peer only to avoid network partitioning
• Selective: Peer with ASes of similar size
• Open: Every co-located AS except customers
![Page 10: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/10.jpg)
10
RESULTS
![Page 11: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/11.jpg)
11
Strategy adoption by transit providers
Conservative Non-conservative0
10
20
30
40
50
60
70
80
90
100
RestrictiveSelectiveOpen
Scenarios
Perc
en
tag
e o
f tr
an
sit
pro
vid
ers
![Page 12: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/12.jpg)
Why do transit providers adopt Open peering?
x y
z w
vSave transit
costs
But your customers are
doing the same!
Affects:• Transit Cost
• Transit Revenue• Peering Cost
![Page 13: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/13.jpg)
Why gravitate towards Open peering?
x y
z wz w,z y,traffic bypasses x
x lost transit revenu
e
Options for x?
x regains lost transit
revenue partially
Y peering openly
x adopts Open peering
Not isolated decisionsNetwork effects !!
z w,traffic passes through x
![Page 14: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/14.jpg)
14
Impact on fitness of transit providers switching from Selective to Open
• 70% providers have their fitness reduced
![Page 15: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/15.jpg)
15
Fitness components: transit cost, transit revenue, peering cost?
• Reduction in transit cost accompanied by loss of transit revenue
![Page 16: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/16.jpg)
Fitness components: transit cost, transit revenue, peering
cost?• Significant increase in peering costs• Interplay between transit & peering cost, transit
revenue
![Page 17: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/17.jpg)
Avoid fitness loss?
x y
z w
x y
z w
vs.
• Lack of coordination• No incentive to unilaterally withdraw from peering
with peer’s customer• Sub-optimal equilibrium
![Page 18: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/18.jpg)
18
Which transit providers gain through Open peering? Which lose?
• Classification of transit providers– Traffic volume– Customer cone size
Small Traffic Small Customers
Small Traffic Large Customers
Large Traffic Small Customers
Large Traffic Large Customers
0
10
20
30
40
50
60
70
80
90
100Strategy adoption by different classes of transit providers
Selective
Open
Restrictive
Perc
enta
ge o
f tr
ansit
pro
vid
ers
in
class
![Page 19: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/19.jpg)
Who loses? Who gains?• Who gains: Small customer cone small traffic volume– Cannot peer with large providers using Selective– Little transit revenue loss
• Who loses: Large customer cone large traffic volume– Can peer with large transit providers with Selective– Customers peer extensively
![Page 20: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/20.jpg)
20
Alternatives: Open peering variants
• Do not peer with immediate customers of peer (NPIC)
• Do not peer with any AS in the customer tree of peer (NPCT)
x y
z wNPIC
x y
z
w
vNPCT
![Page 21: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/21.jpg)
Fitness analysis Open peering variants
• Collective fitness with NPIC approaches Selective
![Page 22: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/22.jpg)
22
• 47% transit providers loose fitness compared to Conservative scheme with Selective strategy
Fitness analysis Open peering variants
OpenNPIC vs. Selective
OpenNPIC vs. Open
![Page 23: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/23.jpg)
23
• Why the improvement?– No traffic “stealing” by peers– Aggregation of peering traffic over fewer
links (economies of scale !!)– Less pressure to adopt Open peering
• Why did collective fitness not increase?– Non-peer transit providers peer openly
with stub customers
Fitness analysis Open peering variants
![Page 24: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/24.jpg)
24
• Why NPIC gives better results than NPCT?– Valley-free routing– x has to rely on provider to reach v
Fitness analysis Open peering variants
24
x y
z
w
vNPCT
![Page 25: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/25.jpg)
25
• Gravitation towards Open peering is a network effect for transit providers
• Extensive Open peering by transit providers in the network results in collective loss
• Coordination required to mitigate• No-peer-immediate-customer can
yield results closer to Selective strategy
Conclusion
![Page 26: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/26.jpg)
26
Thank you
![Page 27: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/27.jpg)
27
Motivation• Traffic ratio requirement for peering? (source: PeeringDB www.peeringdb.com)
![Page 28: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/28.jpg)
28
Introduction
Enterprise customer
Transit Provider
Transit Provider
Internet
Enterprise customer
Content Provider
Content Provider
![Page 29: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/29.jpg)
29
Traffic components Traffic
consumed in the AS
• Transit traffic = Inbound traffic – Consumed trafficsame as
• Transit traffic = Outbound traffic – Generated traffic
Autonomous system
Inbound traffic
Traffic transiting
through the AS
Traffic generated
within the AS
Outbound traffic
![Page 30: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/30.jpg)
30
Customer-Provider traffic comparison – PeeringDB.com
Traffic carried by the AS
![Page 31: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/31.jpg)
31
Customer-Provider traffic comparison
• 2500 customer-provider pairs• 90% pairs: Customer traffic
significantly less than provider traffic• 9.5% pairs: Customer traffic larger
than provider traffic (difference less than 20%)
• 0.05% pairs: Customer traffic significantly larger than provider traffic
![Page 32: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/32.jpg)
32
Geographic presence & constraints
Link formation
across geography
not possible
Regions corresponding to unique
IXPs
Peering link at top tier
possible across regions
Geographic overlap
![Page 33: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/33.jpg)
33
Logical Connectivity
![Page 34: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/34.jpg)
34
Traffic• Traffic for ‘N’’ size network represented
through an N * N matrix• Illustration of traffic matrix for a 4 AS
network
• Consumed traffic
0
0
0
0
30
20
10
030201
t
t
t
ttt
• Generated traffic
Traffic sent by AS 0 to other ASes in
the network
Traffic received by AS 0 from
other ASes in the network
Intra-domain traffic not
captured in the model
N
xii
xiG txV,0
)(
N
xii
ixC txV,0
)(
![Page 35: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/35.jpg)
35
Peering strategy adoption
• Strategy update in each round• Enumerate over all available
strategies• Use netflow to “compute” the fitness
with each strategy• Choose the one which gives maximum
fitness
![Page 36: Analysis of peering strategy adoption by transit providers in the Internet Aemen Lodhi (Georgia Tech) Amogh Dhamdhere (CAIDA) Constantine Dovrolis (Georgia.](https://reader030.fdocuments.us/reader030/viewer/2022032702/56649ce35503460f949ae9b0/html5/thumbnails/36.jpg)
36
Peering strategy adoption
• No coordination• Limited foresight• Eventual fitness can be different • Stubs always use Open peering
strategy
Time
1 2 3
Restrictive Selective Open Restrictive Selective Open Restrictive Selective Open
Open Selective Open