1. Principles in Communication Networks Instructor: Prof. Yuval Shavitt, –Office hours: room 303 s/w eng. bldg., Sun12:00- 13:00 Prerequisites (דרישות קדם): –Introduction to computer communications (TAU, Technion, BGU) Expectations from students: –probability –Queueing theory basics –Graph theory

2. Course Syllabus (tentative) Internet structure Internet measurements Measurement optimization Measurement analysis Introduction to switching, router types Use of Gen. Func.: HOL analysis, TCP analysis. Matching algorithms and their analysis CLOS networks: non-blocking theorem, routing algorithms and their analysis Scheduling algorithms

3. Grade composition Final exam – 60% Project – 30% (Magen) Home assignments (2-3) - 10%

4. Routing in the Internet

5. Routing in the Internet is done in three levels: –In LANs in the MAC layer: Spanning tree protocol for Ethernet Transparent bridge. Source routing for token rings Inside autonomous systems (ASes): –RIP, OSPF, IS-IS, (E)IGRP Between ASes: –BGP

6. Autonomous Systems Autonomous Routing Domains: A collection of physical networks glued together using IP, that have a unified administrative routing policy. An AS is an autonomous routing domain that has been assigned a number. RFC 1930: Guidelines for creation, selection, and registration of an Autonomous System … the administration of an AS appears to other ASes to have a single coherent interior routing plan and presents a consistent picture of what networks are reachable through it.

7. Intra-AS and Inter-AS Routing 1 3 4 3 4 1 2 3 1 2 3 3.1 1.2 2 1 2.1 2 2.2 Both Inter-AS and intra-AS are used to create the routing tables

8. Intra-AS and Inter-AS Routing 1 3 4 3 4 1 2 3 1 2 3 3.1 1.2 2 1 2.1 2 2.2

9. Policy: Inter-AS: admin wants control over how its traffic routed, who routes through its net. Intra-AS: single admin, so no policy decisions needed Scale: hierarchical routing saves table size, reduced update traffic Performance: Intra-AS: can focus on performance Inter-AS: policy may dominate over performance Why different Intra- and Inter-AS routing ?

10. RIP Routing Information Protocol A Distance Vector protocol Based on routed in 4.3 BSD UNIX Distance metric: minimum hop –Max hop = 15 Distance vectors were exchanged every 30 seconds – advertisements Each advertisements route to up to 25 dest networks

11. RIP: Link Failure and Recovery If no advertisement heard for 180 Sec – link declared dead –Route via neighbor invalidated –new advertisements sent to neighbors –neighbors in turn send out new advertisements (if tables changed) –link failure info quickly propagates to entire net –poison reverse used to prevent ping-pong loops (infinite distance = 16 hops)

12. Split Horizon Split Horizon: a node omit from the advertisement any information about destination routed on the link Split Horizon with Poisonous Reverse: All destinations are includes in the message, distance to those routed on the link are set to  Immediately kill two-hop loops

13. RIP-2 Standardized in the 1990s Added authentication RIPng – added support for IPv6

14. Other DV Algorithms IGRP and EIGRP included many improvements Support for multiple metrics –bandwidth, delay, load, MTU, and reliability. EIGRP: included the DUAL algorithm to prevent transient loops

15. OSPF Open Shortest Path First Open – publically available A link state algorithm –Route computation using Dijkstra’s algorithm OSPF advertisement carries one entry per neighbor router Advertisements disseminated to entire AS (via flooding)

16. OSPF Features Security: all OSPF messages authenticated (to prevent malicious intrusion); TCP connections used Multiple same-cost paths allowed For each link, multiple cost metrics for different ToS (e.g., satellite link cost set “low” for best effort; high for real time) Integrated uni- and multicast support: –Multicast OSPF (MOSPF) uses same topology data base as OSPF Hierarchical OSPF in large domains.

17. Hierarchical OSPF Area 1 Area 2 Area 3 backbone Boundary router Area border router Backbone router Internal router

18. Hierarchical OSPF Two-level hierarchy: local area, backbone. –Link-state advertisements only in area –each nodes has detailed area topology; only know direction (shortest path) to nets in other areas. Area border routers: “summarize” distances to nets in own area, advertise to other Area Border routers. Backbone routers: run OSPF routing limited to backbone. Boundary routers: connect to other ASes.

19. IS-IS Intermediate System to Intermediate System A link state protocol Developed in parallel to OSPF Very similar to OSPF Tend to use less messages thus scales better to large networks –ISP grade

20. BGP BGP (Border Gateway Protocol): the de facto standard Path Vector protocol: –similar to Distance Vector protocol –each Border Gateway broadcast to neighbors (peers) entire path (i.e., sequence of ASs) to destination –E.g., Gateway X may send its path to dest. Z: Path (X,Z) = X,Y1,Y2,Y3,…,Z

21. BGP (cont.) Suppose: gateway X send its path to peer gateway W W may or may not select path offered by X –cost, policy (don’t route via competitors AS), loop prevention reasons. If W selects path advertised by X, then: Path (W,Z) = W, Path (X,Z) Note: X can control incoming traffic by controlling its route advertisements to peers: –e.g., if don’t want to route traffic to Z - don’t advertise any routes to Z

22. Why Inter-AS and Intra-AS routing? Policy: Inter-AS: admin wants control over how its traffic routed, who routes through its net. Intra-AS: single admin, so no policy decisions needed Scale: hierarchical routing saves table size, reduced update traffic Performance: Intra-AS: can focus on performance Inter-AS: policy may dominate over performance

23. AS Type of Relationships (ToR) Relationship between a pair of ASes: customer-to-provider relationship provider-to-customer relationship peer-to-peer relationship –E.g., Level3 and AT&T sibling-to-sibling relationship

24. Valley Free Routing If all ASes set their export policies according to the BGP export rules, then an AS path in any BGP routing table entry is valley-free ××× ××××× ××× ×××××

25. A View of the AS Hierarchy Provider - customer

26. A View of the AS Hierarchy No transitivity No SP concatenation Provider - customer Data path

27. A View of the AS Hierarchy Provider - customer Data pathPeer to peer

28. Remarks Since AS connectivity is not always published, certainly ToR is not published While revealing connectivity is not trivial, revealing ToR is very hard.

29. The structure of the Internet

30. How are routers connected? Why should we care? –While communication protocols will work correctly on ANY topology –….they may not be efficient for some topologies –Knowledge of the topology can aid in optimizing protocols

31. The Internet as a graph Remember: the Internet is a collection of networks called autonomous systems (ASs) The Internet graph: –The AS graph Nodes: ASs, links: AS peering –The router level graph Nodes: routers, links: fibers, cables, MW channels, etc. –There are mid-level aggregation schemes PoP topologies, city topologies How does it looks like?

32. Random graphs in Mathematics The Erdös-Rényi model Generation: –create n nodes. –each possible link is added with probability p. Number of links: np If we want to keep the number of links linear, what happen to p as n  ? Poisson distribution

33. The Waxman model Integrating distance with the E-R model Generation –Spread n nodes on a large enough grid. –Pick a link uar and add it with prob. that exponentially decrease with its length –Stop if enough links Heavily used in the 90s

35. 1999 The Faloutsos brothers Measured the Internet AS and router graphs. Mine, she looks different! Notre Dame Looked at complex system graphs: social relationship, actors, neurons, WWW Suggested a dynamic generation model

36. The Faloutsos Graph 1995 Internet router topology 3888 nodes, 5012 edges, =2.57

38. SCIENCE CITATION INDEX (  = 3) Nodes: papers Links: citations (S. Redner, 1998) P(k) ~k -  2212 25 1736 PRL papers (1988) Witten-Sander PRL 1981

39. Sex-web Nodes: people (Females; Males) Links: sexual relationships Liljeros et al. Nature 2001 4781 Swedes; 18-74; 59% response rate.

40. Web power-laws

41. SCALE-FREE NETWORKS (1) The number of nodes (N) is NOT fixed. Networks continuously expand by the addition of new nodes Examples: WWW : addition of new documents Citation : publication of new papers (2) The attachment is NOT uniform. A node is linked with higher probability to a node that already has a large number of links. Examples : WWW : new documents link to well known sites (CNN, YAHOO, NewYork Times, etc) Citation : well cited papers are more likely to be cited again

42. Scale-free model (1) GROWTH : A t every timestep we add a new node with m edges (connected to the nodes already present in the system). (2) PREFERENTIAL ATTACHMENT : The probability Π that a new node will be connected to node i depends on the connectivity k i of that node A.-L.Barabási, R. Albert, Science 286, 509 (1999) P(k) ~k -3

43. The Faloutsos Graph

45. Back to the Internet Understanding its structure and dynamics –help applications (WWW, file sharing) –help improving routing –predict Internet growth So lets look at the data….

46. …Data? The Internet is an engineered system, so someone must know how it is built, no? NO! It is an uncoordinated interconnection of Autonomous Systems (ASes=networks). No central database about Internet structure. Several projects attempt to reveal the structure: Skitter, RouteViews, …

47. The Internet Structure routers

48. The Internet Structure The AS graph

49. Revealing the Internet Structure

52. 30 new links 7 new links NO new links Diminishing return!  Deploying more boxes does not pay-off

53. Revealing the Internet Structure To obtain the ‘ horizontal ’ links we need strong presence in the edge

54. What is DIMES? Distributed Internet measurement and monitoring –Based on software agents downloaded by volunteers Diminishing return? –Software agents  –The cost of the first agent is very high –each additional agent costs almost zero Capabilities –Obtaining Internet maps at all granularity level connectivity, delay, loss, bandwidth, jitter, …. –Tracking the Internet evolution in time –Monitoring the Internet in real time DIMES

55. Diminishing Return? [Chen et al 02], [Bradford et al 01]: when you combine more and more points of view the return diminishes very fast What have they missed? –The mass of the tail is significant No. of views

56. Diminishing Return? [Chen et al 02], [Bradford et al 01]: when you combine more and more points of view the return diminishes very fast What have they missed? –The mass of the tail is significant No. of views

57. Diminish … shminimish

58. How many ASes see an edge? ~9000/6000 are seen only by one

59. Challenges It’s a distributed systems: –Measurement traffic looks malicious Flying under the NOC radar screens (Agents cannot measure too much) –Optimize the architecture: Minimize the number of measurements Expedite the discovery rate BUT agents are –Unreliable –Some move around

60. Agents To be able to use agents wisely we need agents profiles: –Reliablility –Location: Static Bi-homed: where mostly? Mobile: identify home base –Abilities: what type of measurements can it perform? Distributed System complex system real world

61. Agent shavitt Fairly stable measurements from Israel 2 idle weeks Reappear in Spain

64. Degree Distribution k Pr(k) Zipf plot

66. Quantifying the Distribution

67. Data Set Data is obtained from DIMES –Community-based infrastructure, using almost 1000 active measuring software agents –Agents follow a script and perform ~2 probes per minute (ICMP/UDP traceroute, ping) –Most agents measure from a single AS (vp) But some (appear to) measure from more… Data need to be filtered to remove artifacts –Traceroute data collected during March 2008

68. Filtering the data For each agent and each week, classify how many networks it measured the Internet from Typical cases: –AS i :15300, AS j :8 –AS i :10000, AS j :3178 –AS i :10000, AS j :412, AS k :201 –18000, 12, 11, 9, 9, 3, 3, 2, 2, 1, 1, 1, 1, 1, ….

69. Measurements Per Agent Week 4,2008

70. Measurements per Network 500

71. Agents per Network

72. Filtering Results 96% of the agents have less than 4 different vps High degree ASs tend to have more agents High number of measurements for all vps degrees

73. Diminishing Returns? Barford et. al. – the utility of adding many vps quickly diminishes –In terms of ASes and AS-links Shavitt and Shir – utility indeed diminishes but the tail is long and significant –Tail is biased towards horizontal links We wish to quantify how different aspects of AS-level topology are affected by adding more vps

74. Creating topologies per VP sort by

75. Topology Size The return (especially for AS links) does not diminishes fast! VP with small local topology can contribute many new links!

76. Direction of Detected Links For each link: Plot max adjacent AS degree and max adjacent ASes degree difference Low degree difference – indicates tangential links and links between small- size ASes High degree difference – indicates radial links towards the core

77. Convergence of Properties Taking several common AS-level graph properties, and analyze their convergence as local topologies are added –Keeping the sort order by number of links Slow convergence indicates the need to have broad and diverse set of vps

78. Density and Average Degree Slow convergence of density and average degree – easy to detect ASes but difficult to find all links

79. Power-law and Max Degree Fair convergence of power-law exponent Fast convergence of maximal degree – core links are easily detects

80. Betweenness and Clustering Radial links decrease cc Fast convergence of max bc – Level3 (AS3356), a tier-1 AS is immediately detected as having max bc Tangential links increase cc

81. Revisiting Sampling Bias Lakhina et al. – AS degrees inferred from traceroute sampling are biased –ASes in vicinity to vps have higher degrees –Power-law might be an artifact of this! Dall’asta et al. – no…it is quite possible to have unbiased degrees with traceroutes Cohen et al. – when exponent is larger than 2, resulting bias is negligible

82. Evaluating Sampling Bias For each AS find: –All the vps that have it in their local topology –The Valley-Free distance in hops Up-hill to the core (c2p), side-ways in the core (p2p) and down-hill from the core (p2c)

83. Dataset VPs and Distances Low degree ASes are seen from less vps than high-degree ASes…this makes sense! In our dataset, most ASes have a vp that is only 1-2 hops away!

84. Average Distance per Degree Low degree ASes are seen from farther vps…sampling bias? No real bias! More VPs are located in high-degree ASes There are high-degree ASes that are seen from “far” vps Broad distribution – all ASes are pretty close-by to a vp!

85. Revisiting Diversity Bias What is the effect of diversity in vps geo- location and network type? –Some infrastructures rely on academic networks for vp distribution – does it have an effect on the resulting topology? We compare iPlane and DIMES –Classify AS into types: t1,t2, edu, comp, ix, nic using Dimitropoulos et al.

86. Diversity Bias Evaluation iPlane uses many PlanetLab nodes (edu), while DIMES resides mostly at homes (tier-2) Indeed DIMES have higher t2 and comp degrees and iPlane have higher edu degrees – results are slightly biased to vps’ types!

87. In Search of Ground Truth One week is not sufficient for active measurements Both iPlane and DIMES have lower average degrees than RouteViews –Except iPlane’s edu and ix! –Diversity bias exists – need diverse vp types!

88. Measuring Within a Network Comparing vp average degrees to quantify the effect of measuring within a network Indeed, the average degree when measuring within a network is mostly higher (hmm…tier-1 doesn’t count cause most vps are the same!)

89. Conclusion VP distribution is important –Number, AS type, geo-location AS-level graph properties are affected –Some converge very fast –Other converge slowly Community based projects have practically unlimited growth potential!

90. Predicting Growth

91. OurGoal To measure the Internet evolution in time –AS level - too coarse –IP level - too fine

92. The Internet Structure The AS graph

93. The Internet Structure The AS graph The PoP level graph

94. What the PoP is ? PoP – Point of Presence of the ISP

95. OurGoal To measure the Internet evolution in time –AS level - too coarse –IP level - too fine –PoP level – strike the right balance Network size is reasonable Nodes are roughly the same size Has a good geographical grip (with some exceptions) Other uses of PoP maps –Network distance estimation

96. The Algorithm Input & Output

97. Pivot Idea: What is a graph representation of the POP?

98. Comments in 2004 (expert meeting in UCSD) –It will never fly –You’ll be lucky to get 500 downloads in three years –You’ll never be able to clean the noise –How will you deal with problem i (i=1,2,3,4,….)? Status in Feb 2010 –Over 21,700 downloads (over 100 nations) –1000-1200 active agents every day –Measuring from over 200 ASes every week –Data is used world wide by EE, CS, Phys, Econ –DIMES is highly cited DIMES DIMES a historical perspective

100. http://www.netDimes.org