Is Information-Centric Multi-Tree Routing Feasible? ICN workshop - - PowerPoint PPT Presentation

▶
is information centric multi tree routing feasible
SMART_READER_LITE
LIVE PREVIEW

Is Information-Centric Multi-Tree Routing Feasible? ICN workshop - - PowerPoint PPT Presentation

Aug 14, 2022 15 likes •1.04k views

Is Information-Centric Multi-Tree Routing Feasible? ICN workshop 2013 Michele Papalini (University of Lugano) joint work with: Antonio Carzaniga (University of Lugano) Koorosh Khazaei (University of Lugano) Alexander L. Wolf (Imperial College

slide-1
SLIDE 1

Is Information-Centric Multi-Tree Routing Feasible?

ICN workshop 2013 Michele Papalini (University of Lugano)

joint work with:

Antonio Carzaniga (University of Lugano) Koorosh Khazaei (University of Lugano) Alexander L. Wolf (Imperial College London)

slide-2
SLIDE 2

Traditional Networking Information-Centric Networking

slide-3
SLIDE 3

Traditional Networking Information-Centric Networking

addressing end-points addressing information

slide-4
SLIDE 4

Traditional Networking Information-Centric Networking

How do we address information? How do we obtain information?

(Architectural questions)

slide-5
SLIDE 5

Traditional Networking Information-Centric Networking

How do we address information? Tags How do we obtain information? Push/Pull

slide-6
SLIDE 6

Traditional Networking Information-Centric Networking

How do we address information? Tags How do we get information? Push/Pull

Scalable Routing

(System/Evaluation questions)

slide-7
SLIDE 7

How do we address information?

slide-8
SLIDE 8

Solution I: name the data

slide-9
SLIDE 9

Solution I: name the data flat, not human readable identifiers

1DB76EB8DFD6B0B92A293AADC8421830BDE73CB6

hierarchical, meaningful structured names

/ch/usi/inf/papalini/picture.jpg

slide-10
SLIDE 10

/ch/usi/inf/papalini/picture.jpg

slide-11
SLIDE 11

/ch/usi/inf/papalini/picture.jpg /nytimes/sport/baseball/mets /cnn/us/sport/baseball/mets

slide-12
SLIDE 12

/ch/usi/inf/papalini/picture.jpg /nytimes/sport/baseball/mets /cnn/us/sport/baseball/mets /youtube/la dolce vita/HD

slide-13
SLIDE 13

baseball scores for NY Mets

slide-14
SLIDE 14

baseball scores for NY Mets la dolce vita in HD with english subtitles

slide-15
SLIDE 15

Solution II: describe the data

slide-16
SLIDE 16

Solution II: describe the data with set of tags

baseball, new york, mets la dolce vita, HD, en-sub

slide-17
SLIDE 17

Tags

slide-18
SLIDE 18

Tags

more expressivity

/ch/usi/inf/papalini/picture.jpg 1#ch, 2#usi, 3#inf, 4#papalini, 5#picture.jpg

slide-19
SLIDE 19

Tags

more expressivity

/ch/usi/inf/papalini/picture.jpg 1#ch, 2#usi, 3#inf, 4#papalini, 5#picture.jpg

more aggregation

sport, new, football sport Lugano, sport, activities

slide-20
SLIDE 20

Tags

more expressivity

/ch/usi/inf/papalini/picture.jpg 1#ch, 2#usi, 3#inf, 4#papalini, 5#picture.jpg

more aggregation

sport, new, football sport Lugano, sport, activities sport

slide-21
SLIDE 21
slide-22
SLIDE 22

How do we obtain information?

slide-23
SLIDE 23

PULL

slide-24
SLIDE 24

PULL

producer consumer

slide-25
SLIDE 25

PULL

producer consumer forwarding tables

register(descriptors)

slide-26
SLIDE 26

PULL

producer consumer forwarding tables

register(descriptors) interest: content-descriptor

slide-27
SLIDE 27

PULL

producer consumer forwarding tables

register(descriptors) interest: content-descriptor data:content

slide-28
SLIDE 28

PUSH

slide-29
SLIDE 29

PUSH

producer consumer

slide-30
SLIDE 30

PUSH

producer consumer forwarding tables

subscribe(descriptors)

slide-31
SLIDE 31

PUSH

producer consumer forwarding tables

subscribe(descriptors) notification:descriptor

slide-32
SLIDE 32

PU

LL SH

slide-33
SLIDE 33

PU

LL SH

Node A Node B

slide-34
SLIDE 34

PU

LL SH

Node A Node B

register(descriptors)

forwarding tables

slide-35
SLIDE 35

PU

LL SH

Node A Node B

register(descriptors)

forwarding tables

message: content-descriptor +data request: content-descriptor

slide-36
SLIDE 36

PU

LL SH

Node A Node B

register(descriptors)

forwarding tables

message: content-descriptor +data request: content-descriptor reply:data

slide-37
SLIDE 37

PU

LL SH

Node A Node B

register(descriptors)

forwarding tables

message: content-descriptor +data request: content-descriptor reply:data ICN 2011: “Content-Based Publish/Subscribe Networking and Information-Centric Networking”

slide-38
SLIDE 38

PU

LL SH

Node A Node B

register(descriptors)

forwarding tables

message: content-descriptor +data request: content-descriptor reply:data

How do we route?

slide-39
SLIDE 39

Main contribution

Routing schema based on multiple trees

slide-40
SLIDE 40

Main contribution

Routing schema based on multiple trees Scalability analysis

◮ Additional cost (hops/state) using multiple trees ◮ FIBs size using a realistic workload

slide-41
SLIDE 41

Routing

slide-42
SLIDE 42

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c

slide-43
SLIDE 43

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c

slide-44
SLIDE 44

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

next-hop w → predicate Pw

c → pc ∨ pg ∨ ph f → pf ∨ pj ∨ pk e → pa ∨ pd ∨ pe ∨ pi

slide-45
SLIDE 45

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

next-hop w → predicate Pw

c → pc ∨ pg ∨ ph f → pf ∨ pj ∨ pk e → pa ∨ pd ∨ pe ∨ pi g h c

slide-46
SLIDE 46

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c

slide-47
SLIDE 47

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c i j

Stretched Paths

slide-48
SLIDE 48

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c e b

Load

slide-49
SLIDE 49

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c

Multiple Trees

slide-50
SLIDE 50

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

slide-51
SLIDE 51

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ pg ∨ ph T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

Memory Complexity Problem

slide-52
SLIDE 52

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b

slide-53
SLIDE 53

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

slide-54
SLIDE 54

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

slide-55
SLIDE 55

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB)

tree T,next-hop w → predicate PT,w

c → pc ∨ ph ∨ pg T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

slide-56
SLIDE 56

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b

slide-57
SLIDE 57

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk

slide-58
SLIDE 58

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg f → (T1 ∧ pf ) ∨ (T1 ∧ pj) ∨ (T1 ∧ pk) e → (T1 ∧ pa) ∨ (T1 ∧ pd) ∨ (T1 ∧ pe) ∨ (T1 ∧ pi)∨ (T2 ∧ pa) ∨ (T2 ∧ pd) ∨ (T2 ∧ pe) ∨ (T2 ∧ pi)∨ (T2 ∧ pf ) ∨ (T2 ∧ pj) ∨ (T2 ∧ pk)

slide-59
SLIDE 59

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg f → (T1 ∧ pf ) ∨ (T1 ∧ pj) ∨ (T1 ∧ pk) e → (T1 ∧ pa) ∨ (T1 ∧ pd) ∨ (T1 ∧ pe) ∨ (T1 ∧ pi)∨ (T2 ∧ pa) ∨ (T2 ∧ pd) ∨ (T2 ∧ pe) ∨ (T2 ∧ pi)∨ (T2 ∧ pf ) ∨ (T2 ∧ pj) ∨ (T2 ∧ pk)

slide-60
SLIDE 60

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg f → (T1 ∧ pf ) ∨ (T1 ∧ pj) ∨ (T1 ∧ pk) e → pa ∨ pd ∨ pe ∨ pi∨ (T2 ∧ pf ) ∨ (T2 ∧ pj) ∨ (T2 ∧ pk)

slide-61
SLIDE 61

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg f → (T1 ∧ pf ) ∨ (T1 ∧ pj) ∨ (T1 ∧ pk) e → pa ∨ pd ∨ pe ∨ pi∨ (T2 ∧ pf ) ∨ (T2 ∧ pj) ∨ (T2 ∧ pk)

slide-62
SLIDE 62

router b: (FIB)

tree T,next-hop w → predicate PT,w

T1, c → pc ∨ pg ∨ ph T1, f → pf ∨ pj ∨ pk T1, e → pa ∨ pd ∨ pe ∨ pi T2, c → pc ∨ ph ∨ pg T2, e → pa ∨ pd ∨ pe ∨ pf ∨ pi ∨ pj ∨ pk i j k d e f g h a b c b router b: (FIB) next-hop w → predicate PT,w c → pc ∨ ph ∨ pg f → (T1 ∧ pf ) ∨ (T1 ∧ pj) ∨ (T1 ∧ pk) e → pa ∨ pd ∨ pe ∨ pi∨ (T2 ∧ pf ) ∨ (T2 ∧ pj) ∨ (T2 ∧ pk)

coming soon: new data structure

slide-63
SLIDE 63

Evaluation

slide-64
SLIDE 64

Evaluation

Q 1: Is it possible to use trees to route traffic over the Internet?

slide-65
SLIDE 65

Evaluation

Q 1: Is it possible to use trees to route traffic over the Internet? Q 2: Do user-defined descriptor-based addresses aggregate?

slide-66
SLIDE 66

Evaluation

Q 1: Is it possible to use trees to route traffic over the Internet? Q 2: Do user-defined descriptor-based addresses aggregate? We need a workload

slide-67
SLIDE 67

What do we need?

Topology

slide-68
SLIDE 68

What do we need?

Topology Distribute users on the nodes

slide-69
SLIDE 69

What do we need?

Topology Distribute users on the nodes Assign applications to users

slide-70
SLIDE 70

What do we need?

Topology Distribute users on the nodes Assign applications to users Create the registrations

slide-71
SLIDE 71

What do we need?

Topology

◮ AS-level Internet topology

Distribute users on the nodes Assign applications to users Create the registrations

slide-72
SLIDE 72

What do we need?

Topology Distribute users on the nodes

◮ assigned to each AS according to the estimated population

Assign applications to users Create the registrations

slide-73
SLIDE 73

What do we need?

Topology Distribute users on the nodes Assign applications to users

◮ selected according to the real number of users

Create the registrations

slide-74
SLIDE 74

What do we need?

Topology Distribute users on the nodes Assign applications to users Create the registrations

◮ ???

slide-75
SLIDE 75

Imagine the future Internet

◮ study the users behavior on different applications ◮ define registrations with actual tags used by users

slide-76
SLIDE 76

Imagine the future Internet

◮ study the users behavior on different applications ◮ define registrations with actual tags used by users

Push content

◮ web content and blog posts ◮ short messages (tweets)

slide-77
SLIDE 77

Imagine the future Internet

◮ study the users behavior on different applications ◮ define registrations with actual tags used by users

Push content

◮ web content and blog posts ◮ short messages (tweets)

Pull content

◮ videos

slide-78
SLIDE 78

Web Content

Goal: Understand users interests

slide-79
SLIDE 79

Web Content

Goal: Understand users interests Users Bookmarks (Delicious)

◮ bookmarks = subscription

slide-80
SLIDE 80

Application User Registration Delicious 1M 124M Blogs 60K 180K Video 1K 10K Twitter Graph 41M 1B Twitter Messages 400K 500K

slide-81
SLIDE 81

Data Amplification

slide-82
SLIDE 82

Data Amplification

Multiple languages

◮ replicate the data for the 25 most spoken languages ◮ language is chosen according to the popularity

slide-83
SLIDE 83

Data Amplification

Multiple languages

◮ replicate the data for the 25 most spoken languages ◮ language is chosen according to the popularity

Synonyms

◮ for each word we define synonyms ◮ synonyms are randomly chosen

slide-84
SLIDE 84

Evaluation

Q 1: Is it possible to use trees to route traffic over the Internet? Q 2: Do user-defined descriptor-based addresses aggregate?

slide-85
SLIDE 85

Additional cost in using k trees on the actual AS-level topology with k = 8, 16, 32, 64, 128 trees

1 2 3 4 5 6 8 16 32 64 128 Avg/Max Additional Path Length (Hops)

slide-86
SLIDE 86

Additional cost in using k trees on the actual AS-level topology with k = 8, 16, 32, 64, 128 trees

1 2 3 4 5 6 8 16 32 64 128 Avg/Max Additional Path Length (Hops)

slide-87
SLIDE 87

Additional cost in using k trees on the actual AS-level topology with k = 8, 16, 32, 64, 128 trees

1 2 3 4 5 6 8 16 32 64 128 Avg/Max Additional Path Length (Hops)

slide-88
SLIDE 88

Tree aggregation in FIBs with k = 8, 16, 32, 64, 128 trees

2 4 6 8 10 12 14 16 8 16 32 64 128 Distinct Trees per Interface

slide-89
SLIDE 89

Tree aggregation in FIBs with k = 8, 16, 32, 64, 128 trees

2 4 6 8 10 12 14 16 8 16 32 64 128 Distinct Trees per Interface

slide-90
SLIDE 90

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63

slide-91
SLIDE 91

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63

slide-92
SLIDE 92

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63

slide-93
SLIDE 93

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63

slide-94
SLIDE 94

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63 Aggregation Factor 7.85 9.36

slide-95
SLIDE 95

Aggregation of tag-based addresses in FIBs memory requirements in a central node for a single tree 2.5M users All Interfaces Largest Interfaces 325 1 Destinations 42,112 6,559 Tags 276,501,173 35,814,399 Original Descriptors 85,504,514 10,727,593 Actual Descriptors 10,880,657 1,145,713 Size (MB) 518.83 54.63 Bloom Filter size = 400 bits

slide-96
SLIDE 96

Current Work

Workload: 25M users, 513M descriptors, 8 trees Total descriptors: 4.1 billion

slide-97
SLIDE 97

Current Work

Workload: 25M users, 513M descriptors, 8 trees Total descriptors: 4.1 billion Compressed Table: 300M descriptors

slide-98
SLIDE 98

Current Work

Workload: 25M users, 513M descriptors, 8 trees Total descriptors: 4.1 billion Compressed Table: 300M descriptors Update time (average): 6 µsec per descriptor

slide-99
SLIDE 99

Current Work

Workload: 25M users, 513M descriptors, 8 trees Total descriptors: 4.1 billion Compressed Table: 300M descriptors Update time (average): 6 µsec per descriptor Matching time

20 40 60 80 100 120 140 160 180 200 1 2 3 4 5 6 Time (us) Tags

slide-100
SLIDE 100

Conclusion

Routing scheme:

◮ tag-based address ◮ push-pull communication

slide-101
SLIDE 101

Conclusion

Routing scheme:

◮ tag-based address ◮ push-pull communication

Scalability:

◮ AS-topology can be cover with few trees ◮ FIB size is reasonable

slide-102
SLIDE 102

Conclusion

Routing scheme:

◮ tag-based address ◮ push-pull communication

Scalability:

◮ AS-topology can be cover with few trees ◮ FIB size is reasonable

Workload