UNIVERSIT DEGLI STUDI ROMA TRE Dipartimento di Informatica

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione Interdomain Routing Policies in the Internet: Inference and Analysis Massimo Rimondini Defense of the Ph. D in Computer Science and Engineering Mar 16 th, 2007

Interdomain Routing ISP Autonomous Systems BGP WAN LAN Internet

Routing Policies D E I F I S S A L C 1 1 2 2

Why Routing Directions Policies? Interdomain Routing dynamics topology discovery Routing consistency Inference of commercial relationships Oscillation prevention Traffic engineering Debugging Interplay of routing policies Traffic engineering Emulation of computer networks Interconnection strategies ? ?

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione E S N W Interdomain Topology Discovery

State of the Art CAIDA Skitter B. Huffaker, D. Plummer, D. Moore, kc claffy. Topology Discovery by Active Probing. Proc. SAINT ’ 02 kc claffy. Internet Measurement and Data Analysis: Topology, Workload, Performance and Routing Statistics. NAE Workshop, 1999 University of Washington Rocketfuel N. Spring, R. Mahajan, D. Wetherall, T. Anderson. Measuring ISP Topologies with Rocketfuel. IEEE/ACM Trans. on Networking, 2004 N. Spring, R. Mahajan, T. Anderson. Quantifying the Causes of Path Inflation. Proc. ACM SIGCOMM 2003 R. Mahajan, N. Spring, D. Wetherall, T. Anderson. Inferring Link Weights using End-to-End Measurements. Proc. IMW ’ 02

State of the Art AS level Router level

BGP Probing 3 9 9520 5 2 0 520 7 20 0 8 4 1 C 6 C C

BGP Probing 3 9 30{12} 9530{12} 5 2 0 530{12} 7 0{12} 8 4 1 C 6 C C

BGP Probing: Contributions Probing primitives AS-set stuffing Withdrawal observation Exploration algorithm Experimentation on the IPv 4/IPv 6 Internet Route flap dampening analysis Publications Technical report ISCC 2006 Journal (pending acceptance)

BGP Probing: Results Method IPv 6 Links 31 211 (6. 8 ) ASes 24 28 (1. 2 ) Links 23 49 (2. 1 ) 97 (3 ) 222 Level-by-level (7. 2 ) Induced by Withdrawa Protoc W on global l graph ol (I) (W) 29 (1. 2 ) I only 55 (2. 4 ) W only Stable state Withdrawal ASes 32 94 (2. 9 ) IPv 4 IPv 6 312 158 (51%) 175 21 (13%) IPv 6 334 302 168 (50%) 154 (51%) 189 174 23 (14%) 26 (17%)

IRR Data Extraction Internet Routing Registry assignment of Internet resources (IP addresses, AS numbers) routing policies Volunteered updates inconsistent registrations out of date information

IRR Data Extraction Routing Policy Specification Language aut-num: AS 137 import: from AS 20965 action pref=100; from AS 1299 action pref=100; accept ANY [. . . ] export: to AS 1299 announce AS-GARR [. . . ] changed: noc@garr. it 20000830 source: RIPE

IRR Data Extraction Routing Policy Specification Language aut-num: ASX 5 import: { from ASX 2: AS-Z 2 accept 100. 0/8; } refine { from ASX 1 ASX 2 accept 100. 1. 0. 0/16; } except { from ASX 3 accept 100. 1. 1. 0/24; } export: to ASX 1: PRNG-Y 1 to ASX 1: AS-Z 1 except ASX 9 announce 100. 1. 1. 0/24 mp-export: to ASX 11 at 2001: : 1 announce 2001: : /48 default: to ASX 12 action pref=10 default: to ASX 13 100. 1. 1. 1 at 100. 1. 1. 2

IRR Data Extraction aut-num: as-name: descr: import: AS 24336 DIGITALBANK-JP d-b net Backbone from AS 17685 accept ANY export: to AS 17685 announce AS 24336 admin-c: DM 210 -JP tech-c: DM 211 -JP notify: matsuo@po. d-b. ne. jp mnt-by: MAINT-AS 24336 changed: matsuo@po. d-b. ne. jp 20050220 source: RADB aut-num: as-name: descr: AS 24336 DIGITALBANK-JP DIGITALBANK, Inc. , Regional ISP in Japan country: JP import: from AS 17685 action pref=100; accept ANY import: from AS 7682 action pref=100; accept ANY export: to AS 17685 announce AS 24336 export: to AS 7682 announce AS 24336 admin-c: DM 210 -AP tech-c: DM 211 -AP notify: matsuo@po. d-b. ne. jp mnt-routes: MAINT-JP-DIGITALBANK mnt-by: MAINT-JP-DIGITALBANK changed: hm-changed@apnic. net 20050210 source: APNIC

IRR Data Extraction: Contribs A methodology to extract BGP links from the IRR A classification of the links into confidence levels An on-line service Extracted links Statistics about the health of the IRR Collaboration with the RIPE NCC Publications Technical report ACM SIGCOMM Mine. Net 2006

IRR Data Extraction: Results APNI C ARIN APNI C 2688 1 RADB 423 RIPE 19 This work VERI 113 RIPE O RRCC ARI RAD N B 1 423 463 VERI O 19 113 7 14 37 2037 50 45 7 50 1123 23 23666 8 14 45 23 3 310 10852 1 [mahadevan 06] only) 37 RIPE (RIPE 56949

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione Inference of Commercial Relationships E W N

Commercial Relationships 3 9 C 5 2 7 0 8 4 1 6 C C

Commercial Relationships Core 0 1 Peer 6 C 2 4 3 5 Peer Provider Customer 7 8 9 C C

State of the Art X. Dimitropoulos, D. Krioukov, B. Huffaker, kc • combined claffy, G. Riley. Inferring AS Relationships: Dead End or Lively Beginning? LNCS, 2005 approach G. Di Battista, M. Patrignani, M. Pizzonia. Computing • SAT based the Types of the Relationships between Autonomous • heuristics Systems. Proc. IEEE INFOCOM ’ 03 T. Erlebach, A. Hall, T. Schank. Classifying Customer-Provider Relationships in the Internet. Proc. • SAT based • apx algorithms CCN ’ 02 L. Subramanian, S. Agarwal, J. Rexford, R. H. Katz. • combinatorial Characterizing the Internet Hierarchy from Multiple formulation Vantage Points. Proc. IEEE INFOCOM ’ 02 • vantage points L. Gao. On Inferring Autonomous System Relationships in the Internet. IEEE/ACM Transactions on Networking, 2001 Z. Ge, D. R. Figueiredo, S. Jaiswal, L. Gao. On the • hierarchy Hierarchical Structure of the Logical Internet Graph. Proc. SPIE ITCom ’ 01 • policies L. Gao. On Inferring Autonomous System Relationships • in degree based the Internet. Proc. IEEE Global Internet Symposium ’ 00

Validation

Validation Inference algorithms Comparative validation Algorithm independence report Stability report Data set (AS paths)

Validation: Contributions Methodology Set of measures Software suite Validation of [sark 02] against [dpp 03] Publications IPS 2004

Commercial Relationships: Results Algorithm independence (% consistent assignments on same data set) >90% Stability 10 (% consistent assignments over consecutive data sets) # edges [dpp 03]: between 95% and 99% [sark 02]: between 96% and 99% # changes 100

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione Traffic Engineering N S W

Traffic Engineering 2 1 3 0 00 0 0

State of the Art R. K. C. Chang, M. Lo. Inbound Traffic Engineering for Multihomed ASes Using AS Path Prepending. Proc. NOMS ’ 04 B. Quoitin, C. Pelsser, L. Swinnen, O. Bonaventure, S. Uhlig. Interdomain Traffic Engineering with BGP. IEEE Communications Magazine, 2003 N. Feamster, J. Borkenhagen, J. Rexford. Guidelines for Interdomain Traffic Engineering. ACM SIGCOMM Computer Communication Review, 2003 guidelines B. Quoitin, S. Uhlig, C. Pelsser, O. Bonaventure. Internet Traffic Engineering Techniques. TR, 2002 L. Swinnen, S. Tandel, S. Uhlig, B. Quoitin, O. Bonaventure. An Evaluation of BGP-based Traffic Engineering Techniques. TR, 2002 D. Awduche, A. Chiu, A. Elwalid, I. Widjaja, and X. Xiao. Overview and Principles of Internet Traffic Engineering. RFC 3272, 2002. L. Gao, F. Wang. The Extent of observation AS Path Inflation by Routing Policies. Proc. IEEE Global Internet Symposium ’ 02

Seeking for Optimal Prepending ILP formulation Different objective functions Multiple prefixes Tie break Computational geometry ui 1 2 0 X 3 X prepending on ui

Traffic Engineering: Contribs Two formal models Avoid trial-and-error “Pluggable” requirements Algorithmic optimization Publications IPS Mo. Me 2005

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione Interplay of Routing Policies W E S

Policy Probing N. T. Spring. Efficient Discovery of Network Topology and Routing Policy in the Internet. Ph. D Thesis U. Washington. 2004 F. Wang, L. Gao. Inferring and Characterizing Internet Routing Policies. ACM SIGCOMM IM ’ 03 See commercial relationships

Policy Probing 0 1 2 4 7 C 7410 7420. . . 0{1, 6, 7} 0{1, 3, 5, 6, 7, 9} 3 5 8 6 9 10 C C 9630 10 9 52 84 10 10 8 95 42 30 ! ?

Policy Probing: Contributions Path feasibility determination Path preference comparison Experiments on the IPv 4/IPv 6 Internet

Stability J. L. Sobrinho. Network Routing with Path Vector Protocols: Theory and Applications. Proc. ACM SIGCOMM ’ 03 T. G. Griffin, F. Bruce Shepherd, G. Wilfong. The Stable Paths Problem and Interdomain Routing. IEEE/ACM Transactions on Networking, 2002 L. Gao, T. G. Griffin, J. Rexford. Inherently Safe Backup Routing with BGP. Proc. IEEE INFOCOM ’ 01 T. Griffin, G. T. Wilfong. A Safe Path Vector Protocol. Proc. IEEE INFOCOM ’ 00 K. Varadhan, R. Govindan, D. Estrin. Persistent Route Oscillations in Inter-domain Routing. Elsevier Computer Networks, 2000 L. Gao, J. Rexford. Stable Internet Routing without Global Coordination. Proc. ACM SIGMETRICS ’ 00 T. G. Griffin, F. B. Shepherd, G. Wilfong. Policy Disputes in Path-Vector Protocols. Proc. ICNP ’ 99)

Stability 130 10 1 0 210 20 2 3 320 30

Stability: Contributions Event timings Two kinds of stability strict stability forwarding stability Properties Transitions & reachability

UNIVERSITÀ DEGLI STUDI ROMA TRE Dipartimento di Informatica e Automazione E S N W Emulation of Computer Networks

State of the Art Royal Institute of Technology, Sweden IMUNES - An Integrated Multiprotocol Network Emulator / Simulator University of Zagreb Technical University of Madrid UMLMON - Virtualization with User Mode Linux Gerd Stolpmann

Netkit The poor man's system to experiment computer networking Easy to install and use Lightweight Easy setup of complex experiences blah Compact all-in-one virtual labs blah. . . Corpus of ready-to-use labs & teaching material

Contributions Improved user interface Stability Functionalities Usability Improved labs Self tests Lecture slides Publications Technical report Nearly submitted paper F 2. 0 2. 4 2. 0 Beta

Overall. To contributions Do Interdomain topology discovery complexity & effectiveness Commercial relationships of algorithms validate other algorithms Traffic engineering network measurements describe thecomplexity space of solutions bounds on Policy discovery hierarchy case study efficiency Stability game theory accuracy locate stable states Emulation upgrade of networking tools system characterization new features new & improved labs integration with other

Thank you!