Anycast census and geolocation AIMS: Workshop on Active Internet - - PowerPoint PPT Presentation

▶
anycast census and geolocation
SMART_READER_LITE
LIVE PREVIEW

Anycast census and geolocation AIMS: Workshop on Active Internet - - PowerPoint PPT Presentation

Oct 07, 2022 137 likes •340 views

Anycast census and geolocation AIMS: Workshop on Active Internet Measurements 31 March - 2 April 2015 Cicalese Danilo Jordan Auge Diana Joumblatt Dario Rossi Timur Friedman Here is where the anycast instances are Demo: goo.gl/Ff8gdQ 12

slide-1
SLIDE 1

Anycast census and geolocation

AIMS: Workshop on Active Internet Measurements 31 March - 2 April 2015

Cicalese Danilo Jordan Auge Diana Joumblatt Dario Rossi Timur Friedman

slide-2
SLIDE 2

Here is where the anycast instances are

12

Demo: goo.gl/Ff8gdQ

slide-3
SLIDE 3

Here is where the anycast instances are

13

Demo: goo.gl/Ff8gdQ

slide-4
SLIDE 4

14

Unicast geolocation techniques fail with anycast IPs!

Motivation

MaxMind iGreedy MaxMind 8.8.8.8 ?????

slide-5
SLIDE 5

15

Unicast geolocation techniques fail with anycast IPs! Microsoft: 204.79.197.215 iGreedy: 54 instances MaxMind: 1 instance

Motivation

MaxMind iGreedy MaxMind

slide-6
SLIDE 6

16

Unicast geolocation techniques fail with anycast IPs! Microsoft: 204.79.197.215 iGreedy: 54 instances MaxMind: 1 instance Who is using anycast? DNS root server Google DNS Microsoft AT&T CDNs: cloudflare, edgecast

Motivation

iGreedy MaxMind

slide-7
SLIDE 7

Workflow

17

IP targets Blacklist

Ground Truth Public Info

[INFOCOM-15a] Cicalese, Danilo, Joumblatt, Diana , Rossi, Dario, Buob, Marc-Olivier , Auge, Jordan and Friedman, Timur , A Fistful of Pings: Accurate and Lightweight Anycast Enumeration and Geolocation . In IEEE INFOCOM, Hong Kong, China, Apr 2015.

Previous Measurement (Ground Truth)

Active measurement Detection/Geolocation Anycast instances BGP hijack detection Validation Characterization

Ripe Planetlab Ark MLab Dasu … Microsoft

slide-8
SLIDE 8

18

Challenges

slide-9
SLIDE 9

19

Challenges

slide-10
SLIDE 10

110

Challenges

slide-11
SLIDE 11

111

Challenges

slide-12
SLIDE 12

112

Challenges

slide-13
SLIDE 13

113

Challenges

slide-14
SLIDE 14

114

Ripe

The number of vantage points Planetlab O(102), Ripe O(103) How the vantage points are distributed

Planetlab

Recall

slide-15
SLIDE 15

115

Planetlab Ripe

Recall

The number of vantage points Planetlab O(102), Ripe O(103) How the vantage points are distributed Target: 199.27.134.71 CloudFlare Public information: 32 replicas Planetlab: 21 replicas 245 Vantage points 29 Country 186 AS Ripe: 47 replicas 7289 Vantage points 150 Country 2122 AS

slide-16
SLIDE 16

116

Planetlab Ripe Spain France Middle US Nowhere

Precision

slide-17
SLIDE 17

117

Census: 10 millions of targets 1 sample/Vp Probing Rate:1k sample per second Duration: ~ 3 hours

Scalability/Duration

Scanner Scalabity

slide-18
SLIDE 18

118

Census: 10 millions of targets 1 sample/Vp Probing Rate:1k sample per second Duration: ~ 3 hours FastPing:10k sample per second Census: ~ 18 min Lose on recall due the firewalls and filtering

Scalability/Duration

Scanner Scalabity

slide-19
SLIDE 19

PlanetLab X 300 vantage points(VPs) X Limited geographical coverage ~ Accuracy of geolocation ~ Availability issues V Very flexible Very fast What we need: Accurate geolocation of the VP Increase the VP diversity Exploit the complementary of the platform RIPE X 6000 vantage points(VPs)

  • - More constrained(ICMP, traceroute)

X Clean API X Inherently non scalable

How ark can help us

119