Modelling Inter-Domain Routing Olaf Maennel Maennel Olaf - - PowerPoint PPT Presentation
Modelling Inter-Domain Routing Olaf Maennel Maennel Olaf University of Adelaide University of Adelaide Wolfgang Mhlbauer hlbauer Steve Uhlig Wolfgang M Steve Uhlig Technical University Munich Technical University Munich
Steve Uhlig Steve Uhlig
Université catholique de Louvain Université catholique de Louvain
Wolfgang M Wolfgang Mühlbauer ühlbauer
Technical University Munich Technical University Munich
Anja Anja Feldmann Feldmann
Technical University Munich Technical University Munich
Olaf Olaf Maennel Maennel
University of Adelaide University of Adelaide
Matthew Roughan Matthew Roughan
University of Adelaide University of Adelaide
Goal of Goal of
predict an outcome of changes predict an outcome of changes For that we need to know For that we need to know
how routes propagate through the network
which policies are applied on alternative paths Knowing that helps Knowing that helps
to predict impact of policy changes
to predict traffic flows
Understand potential impact of changes in policy: Understand potential impact of changes in policy:
Impact of cancelled peering?
How to identify “ “good good” ” new new peerings peerings? ?
Possibilities to improve policy config config, , e.g.,
e.g.,
check outcome of network configuration network configuration
formulate abstract AS-wide routing policy
Poor path selection, e.g.,
e.g.,
identify highly asymmetric paths
identify long paths
changed connectivity between connectivity between ASes ASes
Predict traffic flow with the help of simulations: Predict traffic flow with the help of simulations:
Inter-domain:
Predict where traffic enters/leaves the network?
How can I balance traffic among my neighbors neighbors? ?
Intra-domain:
How traffic flows within my network?
by simulating end-to-end traceroutes traceroutes
IGP/BGP interactions
Optimize network performance
Locate Internet routing instabilities
Detect problematic routing conditions
Tim Griffin’ ’s BGP wedgies s BGP wedgies
Checking what you are doing:
check if the current state is the desired one
emulate a planned configuration/policy change
real routing information init
real routing information init
real routing information init
real routing information init
Simulation Refinement Model
real routing information
init
Full BGP policies and and decision process decision process
Parse Cisco and and Juniper Juniper configs configs
Parser for Cisco/Juniper router config BGP MRT dumps
Network topology
(routing solver)
configuration
Input Filters Decision Process Output Filters available routes best routes
compute routes for each router towards each prefix
Simulation Refinement Model
real routing information
init
1.
Local-pref 2.
Shortest AS-path 3.
Origin type 4.
Lowest MED value 5.
eBGP over iBGP
6.
Lowest IGP cost (“ “hot-potato hot-potato” ” routing) routing) 7.
Tie-break (e.g., Lowest router-ID)
Simulation Refinement Model
real routing information
init
Split AS, if if multiple paths multiple paths must must be propagated be propagated
Filter, if if longer paths longer paths must must be propagated be propagated
Get rid of random decisions (lowest router-ID lowest router-ID), ), when supporting information is available when supporting information is available
Simulation Refinement Model
real routing information
init
prefix p
prefix p
longer longer path router path router shorter shorter path router path router
prefix p
longer longer path router path router shorter shorter path router path router
prefix p
Filter also on Filter also on “ “egress egress” ”-part of shorter path router.
longer longer path router path router shorter shorter path router path router
=> Fix random decision, if supporting information available => Fix random decision, if supporting information available
lowest lowest neighbor ID neighbor ID decision decision
prefix p
Best match: simulation selects a path that was simulation selects a path that was
RIB-IN: simulation learns a path that was simulation learns a path that was
pick that path as best pick that path as best
not found: No router at the considered AS in No router at the considered AS in the simulation learns about the the simulation learns about the path that was observed in reality path that was observed in reality
training data set: paths that were used to build paths that were used to build model (real observed paths) model (real observed paths)
validation data set: paths that were paths that were unknown unknown to to the model, but that are real the model, but that are real
459 459 1 1 2 2 6 6 28 28 120 120 302 302 validation validation 1391 1391 5 5 10 10 22 22 93 93 359 359 902 902 total total 932 932 total total 600 600 Akamai Akamai 4 4 SpaceNet SpaceNet 8 8 Abilene Abilene 16 16 GEANT GEANT 65 65 RouteViews RouteViews 239 239 RIPE RIPE training training S Source
(selected)
(learned)
(selected)
(learned)
selecting the correct best path
propagate path among possible alternatives
enough observation points
diverse location of observation points
We do not reverse engineer the Internet!
We do not know the real policies!
We “ “only
” have policies which are have policies which are consistent consistent with our observations with our observations… …
intra-domain topology ( (incl
. router configs configs) )
inter-domain topology (incl
. local BGP views) BGP views)
traffic information