1. Inter-domain Routing: Today and Tomorrow Dr. Jia Wang jiawang@research.att.com AT&T Labs Research Florham Park, NJ 07932, USA http://www.research.att.com/~jiawang/ Prof. Zhuoqing Morley Mao zmao@umich.edu Department of EECS University of Michigan Ann Arbor, MI 48109, USA http://www.eecs.umich.edu/~zmao/ IEEE INFOCOM 2004 Tutorial March 8, 2004

2. 2 Outline 1.Overview of Inter-domain routing 2.Routing policies 3.Measuring inter-domain paths 4.Routing instability 5.BGP Beacon - measurement infrastructure 6.Implication on network engineering 7.Security issues Our opinions should not be taken to represent AT&T policies

3. Part I: Overview of Inter- domain Routing

4. March 8, 20044 Internet  Loose cooperative effort of Internet Service Providers (ISPs)  E.g., AT&T, Sprint, UUNet, AOL  Best effort service  Connectedness  Anyone connected to the Internet can exchange traffic with anyone else connected to the Internet

5. March 8, 20045 Internet routing : Routing session routes Internet IP traffic Data plane: forward traffic Control plane: exchange routes rusty.cs.berkeley.edu IP=169.229.62.116 Prefix=169.229.0.0/16 www.cnn.com IP=64.236.16.52 Prefix=64.236.16.0/20

6. March 8, 20046 Internet routing dictates application performance www.cnn.com IP=64.236.16.52 Prefix=64.236.16.0/20 : Routing session routes Internet IP traffic Data plane: forward traffic Control plane: exchange routes Fail over to alternate route rusty.cs.berkeley.edu IP=169.229.62.116 Prefix=169.229.0.0/16

7. March 8, 20047 Internet routing domain  Network devices under same technical and administrative control  Common routing policy  E.g., ISPs, enterprise networks

8. March 8, 20048 Autonomous System (AS)  Autonomous routing domain with an AS number (ASN)  AS numbers  16 bits integer  Public AS number: 1 – 64511  Private AS number: 64512 – 65535  Examples  AT&T: 7018, 6431, …  Sprint: 1239, 1240, …  MIT: 3

9. March 8, 20049 More than 14,000 ASes today QwestSprintUUnet University company AT&T business ISP QwestSprintUUnet University company AT&T business ISP Berkeley Internet CNN Calren Level3 GNN IP traffic Autonomous System Berkeley Calren Level3 Berkeley Calren Level3 QwestSprintUUnet University Company AT&T Business ISP

10. March 8, 200410 Internet Initiative Japan (IIJ)

11. March 8, 200411 IIJ, Tokyo

12. March 8, 200412 Telstra international

13. March 8, 200413 WorldCom (UUNet)

14. March 8, 200414 UUNet, Europe

15. March 8, 200415 Sprint, USA

16. March 8, 200416 AT&T IP Backbone, USA

17. March 8, 200417 GARR-B

18. March 8, 200418 Gigabit research network

19. March 8, 200419 wiscnet.net GO BUCKY!

20. March 8, 200420 MIT.edu http://bgp.lcs.mit.edu/

21. March 8, 200421 Internet routing architecture IP traffic Berkeley CNN Level3 Internet CalrenGNN Inter-domain routing Intra-domain routing

22. March 8, 200422 Intra-domain routing  Run within a certain network infrastructure  Optimize routes taken between points within a network  Internal Gateway Protocols (IGPs)  Metrics based  OSPF (Open Shortest Path First)  RIP (Routing Information Protocol)  IS-IS (Intermediate System to Intermediate System)

23. March 8, 200423 Inter-domain routing  Run between networks  Provide full connectivity of entire Internet  External Gateway Protocol (EBGP)  Policy based  BGP (Border Gateway Protocol)

24. March 8, 200424 Inter-domain routing and BGP  Static routing  Mainly for stub networks  Default routing  Small stub networks  Dynamic routing  Via BGP No need to run BGP in static routing and default routing.

25. March 8, 200425 Link state  Examples: OSPF, IS-IS  Based on Dijkstra’s shortest path computation  Each router periodically floods immediate reachability information to other routers  Fast convergence  High communication and computation overhead  Not scalable for large networks  Requires periodic refreshes

26. March 8, 200426 Vectoring  Distance vs. Path Vector  Distance: hop count (RIP)  Path: entire path (BGP)  Helps identify loops  Supports policy-based routing based on path  Minimal communication overhead  Takes longer to converge, i.e., in proportion to the maximum path length

27. March 8, 200427 Link state vs. vectoring OSPF IS-IS RIP BGP IGP EGP Link stateVectoring BGP is a path vector protocol

28. March 8, 200428 Classful addressing  IPv4: 32 bits  Five classes of networks ClassAddressMask# of networks# of hosts A0*255.0.0.0128~1.6M B10*255.255.0.01638465535 C110*255.255.255.0~2.1M255 DUsed for multicast EReserved and currently unused Improve scaling factor of routing in the Internet => classless

29. March 8, 200429 RFC1519: Classless Inter-domain Routing (CIDR)  No implicit mask based on the class of the network  Explicit masks passed in the routing protocol  Allow aggregation and hierarchical routing

30. March 8, 200430 CIDR addressing 00001100 00100110 00000000 00000000 11111111 11111111 11000000 00000000 IP address: 12.70.0.0Mask: 255.255.252.0 CIDR representation: 12.70.0.0/22 Address Mask Network prefix Host identifier 00001100 00100110 00000000 00000000 11111111 11111111 11000000 00000000

31. March 8, 200431 Address aggregation Internet 12.70.1.0/24 12.70.2.0/24 12.70.3.0/24 12.70.0.0/24 ISP A ISP B 12.70.0.0/22 12.71.0.0/16

32. March 8, 200432 Routing and forwarding  Routing  The decision process of choosing optimal path that is consistent with the administrative or technical policy  Forwarding  The act of receiving a packet, doing a lookup, and copying a packet to the next hop

33. March 8, 200433 Classless forwarding Internet 135.120.0.1 12.70.0.20 IP traffic PrefixNext hop 12.70.0.0/2410.20.0.1 12.70.0.0/1610.20.1.1 12.0.0.0/810.20.128.1 0.0.0.0 10.20.128.10 10.20.0.1 10.20.1.1 10.20.128.1 10.20.128.10

34. March 8, 200434 Inter-domain routing with CIDR support  BGP-4 [RFC1771]  De facto EGP  Path vector protocol  Run on top of TCP for reliability  Carry routing information between ASes  Policy based routing

35. March 8, 200435 BGP basic operations  Set up BGP session  Exchange all candidate routes  Send incremental updates

36. March 8, 200436 Establish BGP session 12.10.0.112.10.0.2 Establish neighboring session between 12.10.0.1 and 12.10.0.2 PrefixNext hop 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 PrefixNext hop 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1 TCP 179

37. March 8, 200437 Exchange all candidate routes 12.10.0.112.10.0.2 PrefixNext hop 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1 PrefixNext hop 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1

38. March 8, 200438 Send incremental updates 12.10.0.112.10.0.2 PrefixNext hop 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1 PrefixNext hop 135.120.0.0/2410.128.0.1 68.35.0.0/1610.192.1.1 12.70.0.0/2410.20.0.1 12.9.0.0/1610.20.1.1 Withdraw 12.9.0.0/16

39. March 8, 200439 BGP messages  OPEN: set up a peering session  UPDATE: announce new routes or withdraw previously announced routes  NOTIFICATION: shut down a peering session  KEEPALIVE: confirm active connection at regular interval

40. March 8, 200440 Internal vs. external BGP Internet I-BGP E-BGP AS A AS B AS C

41. March 8, 200441 I-BGP mesh I-BGP update E-BGP update I-BGP update

42. March 8, 200442 Make I-BGP scale for large AS  Route reflectors  Confederations

43. March 8, 200443 Route reflector E-BGP update RR Only best paths being sent by RR

44. March 8, 200444 Confederation AS 1000 EBGP IBGP AS 65010 AS 65020

45. March 8, 200445 BGP updates  Three blocks  Prefix  Path attributes  Unreachable routes

46. March 8, 200446 BGP attributes Value Code Reference 1 ORIGIN [RFC1771] 2 AS_PATH [RFC1771] 3 NEXT_HOP [RFC1771] 4 MULTI_EXIT_DISC [RFC1771] 5 LOCAL_PREF [RFC1771] 6 ATOMIC_AGGREGATE [RFC1771] 7 AGGREGATOR [RFC1771] 8 COMMUNITY [RFC1997] 9 ORIGINATOR_ID [RFC1998] 10 CLUSTER_LIST [RFC1998] 11 DPA [Chen] 12 ADVERTISER [RFC1863] 13 RCID_PATH / CLUSTER_ID [RFC1863] 14 MP_REACH_NLRI [RFC2283] 15 MP_UNREACH_NLRI [RFC2283] 16 EXTENDED COMMUNITIES [Rosen] 17 NEW_AS_PATH [E.Chen] 18 NEW_AGGREGATOR [E.Chen] 19 SAFI Specific Attribute (SSA) [Nalawade] 20-254 Unassigned 255 reserved for development http://www.iana.org/assignments/bgp-parameters

47. March 8, 200447 Establish connectivity 135.120.0.0/16 12.10.0.1 12.10.0.2 PrefixNext hop AS path 135.120.0.0/1612.10.0.11 EBGP IBGP EBGP 12.10.0.5 12.10.0.6 AS 1AS 2 AS 3 PrefixNext hop AS path 135.120.0.0/1612.10.0.52 1 PrefixNext hop AS path 135.120.0.0/1612.10.0.11

48. March 8, 200448 IGP and BGP working together 135.120.0.0/16 12.10.0.1 12.10.0.2 PrefixNext hop AS path 135.120.0.0/1612.10.0.11 EBGP IBGP EBGP 12.10.0.5 12.10.0.6 AS 1 AS 2 AS 3 PrefixNext hop AS path 135.120.0.0/1612.10.0.11 10.10.0.1 PrefixNext hop 12.10.0.0/3010.10.0.1 135.120.0.0/1610.10.0.1 12.10.0.0/30