DISSECTING DNS STAKEHOLDERS IN MOBILE NETWORKS 2 CoNEXT 2017, - - PowerPoint PPT Presentation

▶
dissecting dns stakeholders in mobile networks
SMART_READER_LITE
LIVE PREVIEW

DISSECTING DNS STAKEHOLDERS IN MOBILE NETWORKS 2 CoNEXT 2017, - - PowerPoint PPT Presentation

Sep 12, 2023 188 likes •476 views

Mario Almeida, Alessandro Finamore, Diego Perino, Narseo Vallina-Rodriguez, Matteo Varvello CoNEXT 2017 - Seoul/Incheon, South Korea DISSECTING DNS STAKEHOLDERS IN MOBILE NETWORKS 2 CoNEXT 2017, Seoul/Incheon WHY TO STUDY DNS IN MOBILE

slide-1
SLIDE 1

DISSECTING DNS STAKEHOLDERS
 IN MOBILE NETWORKS

CoNEXT 2017 - Seoul/Incheon, South Korea

Mario Almeida, Alessandro Finamore, Diego Perino, Narseo Vallina-Rodriguez, Matteo Varvello

slide-2
SLIDE 2
  • Complex scenario as domain owners, operators, app developers, and OSes
  • perate autonomously
  • DNS is prominent in mobile traffic, up to 50% of all flows [1]
  • Performance wise, only query resolution time level has been considered [2,3]

CoNEXT 2017, Seoul/Incheon

WHY TO STUDY DNS IN MOBILE NETWORKS?

[1] “Application Bandwidth and Flow Rates from 3 Trillion Flows Across 45 Carrier Networks” PAM’17
 [2] “QoE Doctor: Diagnosing Mobile App QoE with Automated UI Control and Cross-layer Analysis” IMC’14
 [3] “Behind the Curtain: Cellular DNS and Content Replica Selection” IMC’14

2

slide-3
SLIDE 3

WHY TO STUDY DNS IN MOBILE NETWORKS?

  • Complex scenario as domain owners, operators, app developers, and OSes
  • perate autonomously
  • DNS is prominent in mobile traffic, up to 50% of all flows [1]
  • Performance wise, only query resolution time level has been considered [2,3]

[1] “Application Bandwidth and Flow Rates from 3 Trillion Flows Across 45 Carrier Networks” PAM’17
 [2] “QoE Doctor: Diagnosing Mobile App QoE with Automated UI Control and Cross-layer Analysis” IMC’14
 [3] “Behind the Curtain: Cellular DNS and Content Replica Selection” IMC’14

3

  • Who is responsible for all this traffic?
  • Is it really needed?
  • What is the role of DNS on users QoE?

QUESTIONS

CoNEXT 2017, Seoul/Incheon

slide-4
SLIDE 4

MOBILE NETWORKS DNS STAKEHOLDERS

4

STAKEHOLDERS DNS COMPONENT

Domain properties propagation

Domain owners & CDNs

ADNS

Authoritative DNS resolver

Control domain properties:

  • domain-to-IPs mapping
  • time to live (TTL)

Mobile Network Operators

MNOs

LDNS

Local recursive DNS resolver

Handle devices queries:

  • Serves cached ADNS data
  • Recursively query ADNS
  • Can overwrite ADNS data


(TTL violations)

Developers
 & OSes

cDNS

On-device client DNS resolver

Local cache:

  • Controlled by the OS
  • Developers can bypass it 


using raw sockets

FUNCTION

CoNEXT 2017, Seoul/Incheon

slide-5
SLIDE 5

MOBILE NETWORKS DNS STAKEHOLDERS

5

STAKEHOLDERS DNS COMPONENT

Domain properties propagation

Domain owners & CDNs

ADNS

Authoritative DNS resolver

Control domain properties:

  • domain-to-IPs mapping
  • time to live (TTL)

Mobile Network Operators

MNOs

LDNS

Local recursive DNS resolver

Handle devices queries:

  • Serves cached ADNS data
  • Recursively query ADNS
  • Can overwrite ADNS data


(TTL violations) Local cache:

  • Controlled by the OS
  • Developers can bypass it 


using raw sockets

FUNCTION

CoNEXT 2017, Seoul/Incheon

EACH STAKEHOLDER PLAYS AN IMPORTANT ROLE

Developers
 & OSes

cDNS

On-device client DNS resolver

slide-6
SLIDE 6

DATASETS

6

LDNS ADNS

CoNEXT 2017, Seoul/Incheon

cDNS

slide-7
SLIDE 7

Name Type Dur Apps User Domains Flows IPs MNO 1M

  • 19M

198M 250M 4.2 Lumen 1.5Y 8,279 5k 35k 5.3M 99k NexusTTL 1M host 1 10k 104k 20k NexusPLT 1M chrome 1 6k 46k 8k

DATASETS

7

LDNS ADNS Webproxy

Operator network

CoNEXT 2017, Seoul/Incheon

AD-HOC PROBING AD-HOC PROBING IN-NETWORK

(*) https://play.google.com/store/apps/details?id=edu.berkeley.icsi.haystack&hl=en

ON-DEVICE

cDNS

slide-8
SLIDE 8

Name Type Dur Apps User Domains Flows IPs MNO 1M

  • 19M

198M 250M 4.2 Lumen 1.5Y 8,279 5k 35k 5.3M 99k NexusTTL 1M host 1 10k 104k 20k NexusPLT 1M chrome 1 6k 46k 8k

DATASETS

8

LDNS ADNS Webproxy

Operator network

CoNEXT 2017, Seoul/Incheon

IN-NETWORK ON-DEVICE AD-HOC PROBING AD-HOC PROBING

(*) https://play.google.com/store/apps/details?id=edu.berkeley.icsi.haystack&hl=en cDNS

slide-9
SLIDE 9

DATASETS

Name Type Dur Apps User Domains Flows IPs MNO 1M

  • 19M

198M 250M 4.2 Lumen 1.5Y 8,279 5k 35k 5.3M 99k NexusTTL 1M host 1 10k 104k 20k NexusPLT 1M chrome 1 6k 46k 8k

9

LDNS ADNS Webproxy Alexa

Operator network

CoNEXT 2017, Seoul/Incheon

IN-NETWORK ON-DEVICE AD-HOC PROBING AD-HOC PROBING

(*) https://play.google.com/store/apps/details?id=edu.berkeley.icsi.haystack&hl=en cDNS

slide-10
SLIDE 10

DATASETS

10

LDNS ADNS Webproxy Alexa

Operator network

CoNEXT 2017, Seoul/Incheon

IN-NETWORK ON-DEVICE AD-HOC PROBING AD-HOC PROBING

+

Top-1M to 
 compare popul. 20k apps for
 static analysis

Name Type Dur Apps User Domains Flows IPs MNO 1M

  • 19M

198M 250M 4.2 Lumen 1.5Y 8,279 5k 35k 5.3M 99k NexusTTL 1M host 1 10k 104k 20k NexusPLT 1M chrome 1 6k 46k 8k

(*) https://play.google.com/store/apps/details?id=edu.berkeley.icsi.haystack&hl=en cDNS

slide-11
SLIDE 11

ANALISYS ROADMAP

Domains Footprint Domain Properties Configs & Apps Design Impact on QoE

  • What are the relevant domains?
  • What the role of the OS?
  • What the role of Apps?
  • Original values at the ADNS
  • How LDNS cache/mingle 


those properties

  • On-device caching performance
  • Are explicit proxies widely adopted?
  • Are developers using OS configurations?
  • DNS impact on webpage 


page load time (PLT)

11

CoNEXT 2017, Seoul/Incheon

slide-12
SLIDE 12

ANALISYS ROADMAP

Domains Footprint Domain Properties Configs & Apps Design Impact on QoE

  • What are the relevant domains?
  • What the role of the OS?
  • What the role of Apps?
  • Original values at the ADNS
  • How LDNS cache/mingle 


those properties

  • On-device caching performance
  • Are explicit proxies widely adopted?
  • Are developers using OS configurations?
  • DNS impact on webpage 


page load time (PLT)

12

CoNEXT 2017, Seoul/Incheon

Selection Selection Selection Selection

slide-13
SLIDE 13
  • 198M domains in MNO dataset, but top-10k most popular generate 87% flows

13

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON POPULAR DOMAINS

slide-14
SLIDE 14
  • 198M domains in MNO dataset, but top-10k most popular generate 87% flows

14

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON POPULAR DOMAINS

POPULAR DOMAINS DRIVE FLOWS COUNT

slide-15
SLIDE 15
  • 198M domains in MNO dataset, but top-10k most popular generate 87% flows

15

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON POPULAR DOMAINS

POPULAR DOMAINS DRIVE FLOWS COUNT BECAUSE THEY ARE ALSO
 COMMON ACROSS APPS

slide-16
SLIDE 16
  • Out of 198M, 162M (82%) domains are used only once in 1 month

16

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON UNPOPULAR DOMAINS

slide-17
SLIDE 17
  • Out of 198M, 162M (82%) domains are used only once in 1 month

17

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON UNPOPULAR DOMAINS

UNPOPULAR DOMAINS 
 EPHEMERAL

example d-2294771243204135673.ampproject.net

slide-18
SLIDE 18
  • Out of 198M, 162M (82%) domains are used only once in 1 month

18

CoNEXT 2017, Seoul/Incheon

DOMAINS FOOTPRINT: FOCUS ON UNPOPULAR DOMAINS

UNPOPULAR DOMAINS 
 EPHEMERAL TRACKING/PERSONALIZATION
 INTRODUCES OVERHEAD

example d-2294771243204135673.ampproject.net

5 services handle 80% of ephemeral domains

slide-19
SLIDE 19

TTL POLICIES ARE AGGRESSIVE

19

CoNEXT 2017, Seoul/Incheon

▸ 50% of domains have TTL < 60s

slide-20
SLIDE 20

TTL POLICIES ARE AGGRESSIVE

20

CoNEXT 2017, Seoul/Incheon

▸ 50% of domains have TTL < 60s ▸ This impacts on-device caching performance

Simulation based on domains 
 requested more than once

slide-21
SLIDE 21

TTL POLICIES ARE AGGRESSIVE

21

CoNEXT 2017, Seoul/Incheon

▸ 50% of domains have TTL < 60s ▸ This impacts on-device caching performance

Simulation based on domains
 requested more than once

slide-22
SLIDE 22

DNS IMPACT ON WEBPAGES PLT

CoNEXT 2017, Seoul/Incheon

22

▸ Consider top-1k Alexa pages, and measure DNS latency over the critical path


(i.e., content downloaded entirely/partially in isolation)

slide-23
SLIDE 23

DNS IMPACT ON WEBPAGES PLT

CoNEXT 2017, Seoul/Incheon

23

▸ Consider top-1k Alexa pages, and measure DNS latency over the critical path


(i.e., content downloaded entirely/partially in isolation)

slide-24
SLIDE 24

QUICK OVERVIEW OF OTHER RESULTS

  • Alexa rank does not well intersect with the popular domains
  • iOS and Android share popular domains, but iOS devices are more “chatty”
  • Aggressive TTL values, but domains have <10 IPs over 1 month
  • Almost no TTL violations found, but LDNS architecture can impact caching

performance

  • Explicit proxies are not widely adopted, nor developer bypass OS config

CoNEXT 2017, Seoul/Incheon

slide-25
SLIDE 25

…SO DNS HAS AN IMPACT HOW DO WE REDUCE IT?

slide-26
SLIDE 26

DESIGN OPTIONS

Ideally one would like not to have any DNS traffic

CoNEXT 2017, Seoul/Incheon

26

Name Popular Stakeholder Pros Cons Explicit proxy No Operator No DNS on radio access From tests, reduces only 50% DNS latency on PLT Domains pre-fetching No Developer Lower latency More DNS traffic Domains pre-staging

  • OS/Operators

From tests, is the best 
 performing Complex to engineer

slide-27
SLIDE 27

GOING BEYOND THIS PRELIMINARY WORK

CoNEXT 2017, Seoul/Incheon

27

What is the “PLT”

  • f generic 


mobile apps
 traffic?

What is on the 
 “critical path”
 beyond DNS?

slide-28
SLIDE 28

THANK

YOU!

?

…