Route Aggregation on the Global Network DRAGON João Luís Sobrinho 1 Laurent Vanbever 2 Franck Le 3 Jennifer Rexford 2 1 Instituto Telecomunicações 1 Universidade de Lisboa ID: 239784
Download Presentation The PPT/PDF document "Distributed" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Distributed Route Aggregation on the Global Network(DRAGON)
João Luís Sobrinho1
Laurent Vanbever
2, Franck Le3, Jennifer Rexford2
1
Instituto Telecomunicações,
1
Universidade de Lisboa
2
Princeton
University
,
3
IBM T. J. Watson Research Slide2
Recently in the news (August 2014)
+512 K
IPv4 prefixes propagated to ≈50.000 ASs 2Slide3
Not a scalable Internet routing systemMost prefixes propagated (by BGP) to all
ASsRouting & forwarding tables growth
Churn & convergence time increaseS*BGP processing requirements escalation3Slide4
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategyDRAGON: additional aspects
DRAGON: performanceConclusions4Slide5
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategy
DRAGON: additional aspectsDRAGON: performanceConclusions
5Slide6
Decentralization: each AS decides…Where to acquire address
space provider? provider-independent, Internet registry?
Where to connectmulti-homing? peering at an exchange point?How to announce assigned address space de-aggregate first?How to treat routes learned from neighborswhich routing policies?6Slide7
Structure: opportunities to scale?
Hierarchy
: IP prefixes
Classless Inter Domain Routing7Slide8
Structure: opportunities to scale?Provider-customer agreements
Hierarchy
: ASs
Hierarchy: IP prefixes
Classless
Inter
Domain
Routing
8Slide9
Structure: opportunities to scale?Provider-customer agreements
Hierarchy
: ASs
Hierarchy: IP prefixes
Classless
Inter
Domain
Routing
9Slide10
Structure: opportunities to scale?Provider-customer agreements
Hierarchy
: ASs
Hierarchy: IP prefixes
Geography
(
rough
)
Classless
Inter
Domain
Routing
10Slide11
Structure: opportunities to scale?Provider-customer agreements
How
to exploit this
structure for scalability?
Hierarchy
:
ASs
Hierarchy
:
IP prefixes
Geography
(
rough
)
Classless
Inter
Domain
Routing
11Slide12
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategy
DRAGON: additional aspectsDRAGON: performanceConclusions
12Slide13
Filtering strategyFilter the more specific prefixes when
possibleno black holes
strive to preserve global forwarding behavior Use incentives to filter locallysave on routing and forwarding stateforward data-packets along best possible routeMake standard usage of BGP routing messages13Slide14
Generation of aggregation prefixesGenerate aggregation prefixes when
beneficialpermit filtering of provider-independent
prefixesnew address space is not createdAnnounce as in BGPself-organization when more than one AS generates the same aggregation prefix14Slide15
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategy
DRAGON: additional aspectsDRAGON: performanceConclusions
15Slide16
Providers, customers, and peers
#1
#2
#3
#4
#5
#6
#7
#8
#9
AS
peer
peer
provider
customer
16Slide17
Prefixesq more specific than p
originates p (10.0.0.0/16)
originates q (10.0.0.0/24)
#1
#2
#3
#4
#5
#6
#7
#8
#9
17Slide18
BGP: Gao-Rexford routing policies
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
exportation
:
all
routes
from
customers
all
routes
to
customers
preference
q
q
-
route
(
route
pertaining
to
q
)
#1
#2
#3
#4
#5
#6
#7
#8
#9
+
“
peer
”
18
– Slide19
BGP: Gao-Rexford routing policies
q
#1
#2
#3
#4
#5
#6
#7
#8
#9
q
-
route
exportation
:
all
routes
from
customers
all
routes
to
customers
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
–
“
peer
”
19Slide20
BGP: Gao-Rexford routing policies
q
#1
#2
#3
#4
#5
#6
#7
#8
#9
q
-
route
exportation
:
all
routes
from
customers
all
routes
to
customers
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
“
peer
”
–
20Slide21
BGP: Gao-Rexford routing policies
q
#1
#2
#3
#4
#5
#6
#7
#8
#9
q
-
route
exportation
:
all
routes
from
customers
all
routes
to
customers
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
“
peer
”
–
21Slide22
Final state for prefix q
q
#1
#2
#3
#4
#5
#6
#7
#8
#9
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
“
peer
”
–
22Slide23
Final state for prefix p
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
“
peer
”
–
23Slide24
Combined
states
for q and p
p
-
route
q
-
route
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
forwarding
:
longest
prefix
match rule
q
route
attributes
:
“
learned
from
…”
“
customer
”
“
provider
”
preference
+
“
peer
”
–
24Slide25
Filtering Code (FC)Filtering Code (FC)
Other than the owner
of p, in the presence of p, filter q if only if:attribute of p-routesame or preferred toattribute of q-route 25Slide26
Filtering
Code
(FC)
Filtering
Code
(FC)
Other
than
the
owner
of
p
, in
the
presence
of
p,
filter
q
if
only
if
:
attribute
of
p
-
route
same
or
preferred
to
a
ttribute
of
q
-
route
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
AS
1,
AS
2,
AS
3, AS 4, AS 5, AS 8
filter
q
on
executing
the
FC ( )
26Slide27
Arbitrary
AS
applies
the FC
AS 4
applies
the
FC
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
27
withdrawal
of
q
-
routeSlide28
Arbitrary
AS
applies
the
FC
AS 4
applies
the
FC
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
28
attribute
of
q
-
route
worsens
at
AS 3:
double
incentive to
apply
the
FC ( )
saves
on
forwarding
state
restores
attribute
of
route
used
to
forward
data-
packets
with
destination
in
qSlide29
Neighbor
AS
applies
the
FC
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
29
AS 3
applies
the
FCSlide30
All
ASs
apply
the
FC
AS 6, AS 7, AS 9
detailed
information
q
AS 1, AS 2, AS3, AS 4, AS 5,
AS 8
coarse-grained
information
p
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
30Slide31
Global
property
:
c
orrectness
Correctness
no
routing
anomalies
(no
black
holes
)
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
31Slide32
Global
property
:
r
oute
consistency
Route
consistency
attribute
of
route
used
to
forward
data-
packets
is
preserved
Optimal
r
oute
consistency
set
of
ASs
that
forgo
q
is
maximal for
route
consistency
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
32Slide33
Route
consistency
:
p
artial
deployment
AS
5
, AS 8
filter
q
r
oute
consistency
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
33Slide34
Route
consistency
:
p
artial
deployment
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
AS
5
, AS 8
filter
q
r
oute
consistency
AS 1, AS 2
filter
q
route
consistency
34Slide35
Route
consistency
:
p
artial
deployment
AS
5
, AS 8
filter
q
r
oute
consistency
AS 1, AS 2
filter
q
route
consistency
AS 3
filters
q
route
consistency
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
35Slide36
Route
consistency
:
p
artial
deployment
AS
5
, AS 8
filter
q
r
oute
consistency
AS 1, AS 2
filter
q
route
consistency
AS 3
filters
q
route
consistency
AS
4
filters
q
route
consistency
#1
#2
#3
#4
#5
#6
#7
#8
#9
p
q
36Slide37
Filtering strategy: general caseCorrectnessfor all
routing policies for which BGP is correct
Route consistent states culminating in optimalityfor isotone routing policies (includes Gao-Rexford)otherwise, some stretchOptimal route consistency is not synonymous with efficiency (think shortest paths)37Slide38
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategy
DRAGON: additional aspectsDRAGON: performanceConclusions
38Slide39
Additional aspects of DRAGONPrefixes at
multiple levels of specificityparent
prefix and child prefixes39Slide40
Additional aspects of DRAGONPrefixes at
multiple levels of specificityparent
prefix and child prefixesGeneration of aggregation prefixespermit filtering of provider-independent prefixes 40Slide41
Additional aspects of DRAGONPrefixes at
multiple levels of specificityparent
prefix and child prefixesGeneration of aggregation prefixespermit filtering of provider-independent prefixes Network dynamicsadapts to link failures and additions41Slide42
OutlineCharacterizing the Internet for scalabilityDRAGON: basic ideasDRAGON: filtering strategy
DRAGON: additional aspectsDRAGON: performanceConclusions
42Slide43
Filtering efficiency# (FIB entries BGP) – # (FIB entries DRAGON)
# (FIB
entries BGP)50% of the prefixes without parentFiltering efficiency bounded at 50%Bound on filtering efficiency rises to 79%
Current set of prefixesWith aggregation prefixes43Slide44
Performance of DRAGON
Every
AS forgoes at least 47.5% of the prefixes80% ASs realize the maximum filtering efficiency of 50% 80%
ASs realize the maximum filtering efficiency of 79% Every AS forgoes at least 70% of the prefixesFIB aggregationcumulated% ASsfiltering
efficiency
current
set
of
prefixes
with
aggregation
prefixes
44Slide45
ConclusionsDRAGON is a BGP add-on to scale the Internet
routing systemDRAGON can be deployed
incrementallyDRAGON can reduce the amount of state in the Internet routing system by approximately 80%DRAGON is – more fundamentally – a solid framework to reason about route aggregation45Slide46
Thank you!
Visit us at www.route-aggregation.net
46