Neutral Net Neutrality Expressing User Preferences with Network - - PowerPoint PPT Presentation

▶
neutral net neutrality
SMART_READER_LITE
LIVE PREVIEW

Neutral Net Neutrality Expressing User Preferences with Network - - PowerPoint PPT Presentation

May 15, 2023 270 likes •769 views

Neutral Net Neutrality Expressing User Preferences with Network Cookies Yiannis Yiakoumis, Sachin Katti, Nick McKeown Stanford University Who controls your network access? 2 Who controls your network access? 3 Who controls your network

slide-1
SLIDE 1

Neutral Net Neutrality

Expressing User Preferences with Network Cookies Yiannis Yiakoumis, Sachin Katti, Nick McKeown Stanford University

slide-2
SLIDE 2

2

Who controls your network access?

slide-3
SLIDE 3

Who controls your network access?

3

slide-4
SLIDE 4

Who controls your network access?

Neutral Network ISP-defined Fast Lanes

What about the users?

4

slide-5
SLIDE 5

What if we let users decide?

User-Driven

5

Neutral Network ISP-defined Fast Lanes

slide-6
SLIDE 6

Outline

  • Why user preferences matter
  • Expressing user preferences to the network
slide-7
SLIDE 7

Outline

  • Why user preferences matter
  • Expressing user preferences to the network
slide-8
SLIDE 8
  • 1. Boost : User-driven Fast Lane

Internet

8

slide-9
SLIDE 9
  • Deployed in ~300 homes (Google employees)
  • One Boost lasts for one hour + Last Boost wins
  • 44 websites in two weeks

9

  • 1. Boost : User-driven Fast Lane
slide-10
SLIDE 10

How do people use Boost?

  • “Business-related stuff should be faster (me or

my partner)”

  • “I am stuck with a 2Mbps connection, and very
  • ften I need to dedicate all of it on a single task.”
  • “I want my Netflix movie to go faster, not my

daughter’s Netflix” User context

10

slide-11
SLIDE 11

1 10 100 1000 >5000 mail.google.com www.netflix.com www.nbc.com www.cnn.com www.facebook.com abc.go.com speedtest.net www.usanetwork.com www.cucirca.eu www.hulu.com www.ticketmaster.com www.espncricinfo.com www.intercallonline.com www.ondemandkorea.com www.starsports.com www.skai.gr 2 4 6 8 10 # of users

How do people use Boost?

11

slide-12
SLIDE 12

1 10 100 1000 >5000 Alexa ranking mail.google.com www.netflix.com www.nbc.com www.cnn.com www.facebook.com abc.go.com speedtest.net www.usanetwork.com www.cucirca.eu www.hulu.com www.ticketmaster.com www.espncricinfo.com www.intercallonline.com www.ondemandkorea.com www.starsports.com www.skai.gr 2 4 6 8 10 # of users

How do people use Boost?

12

Heavy-tail of user preferences

slide-13
SLIDE 13
  • 2. Zero-rating

13

slide-14
SLIDE 14

Do users want zero-rating?

Interested 63.0% Don't care Don't know 37.0%

14

1000 smartphone owners, US, 18-65 year old, SurveyMonkey Audience, 08/15

slide-15
SLIDE 15

facebook netflix instagram google maps snapchat google music whatsapp reddit is fun amazon music nine cnn.com wikipedia tunein radio beats hulu nyt trivia crack candy crush flipboard viber soma.fm swig indie103.1 lynda.com schwab 8tracks edmodo mapmyrun action news wwf 1 5 10 15 20 # of users

~

~50

What do users really want to zero-rate?

Which application would you choose to zero-rate?

15

Heavy-tail of user preferences

slide-16
SLIDE 16

facebook netflix instagram google maps snapchat google music whatsapp reddit is fun amazon music nine cnn.com wikipedia tunein radio beats hulu nyt trivia crack candy crush flipboard viber soma.fm swig indie103.1 lynda.com schwab 8tracks edmodo mapmyrun action news wwf 1 5 10 15 20 # of users

~

~50

What do users really want to zero-rate?

Which application would you choose to zero-rate?

16

80% ignored by all programs!

slide-17
SLIDE 17
  • Users have unique and diverse preferences
  • Respect the heavy-tail
  • Let users decide

User Preferences Takeaways

17

slide-18
SLIDE 18

Outline

  • Why user preferences matter
  • Expressing user preferences to the network

– Why existing mechanisms don’t work – Network Cookies

18

slide-19
SLIDE 19

Deep Packet Inspection

Internet Boost cnn.com

What is cnn.com ???

19

cnn.com skai.gr 20 40 60 80 100 packets boosted (%)

matched

DPI is not expressive

slide-20
SLIDE 20

DiffServ

Internet

  • 64 predefined classes
  • No authentication
  • No revocation

20

DiffServ doesn’t work across network boundaries

slide-21
SLIDE 21

“I want traffic X to get service Y!”

  • 1. I want Netflix to be faster
  • 2. I want Spotify zero-rated
  • 3. I want low-latency Skype for work

Internet

21

slide-22
SLIDE 22

“I want traffic X to get service Y!”

Internet

  • 1. I want Netflix to be faster
  • 2. I want Spotify zero-rated
  • 3. I want low-latency Skype for work

22

  • 3. Configure the network
  • 2. Communicate it to

the network

slide-23
SLIDE 23

Free Lane

How do we map traffic to a lane?

23

slide-24
SLIDE 24

Network Cookies: a mapping abstraction

Free Lane Network Cookie : A small piece of data users append to their traffic

  • 1. Get cookie for each service
  • 2. User appends cookies to the desired traffic
  • 3. Network matches against them and enforces service

24

slide-25
SLIDE 25

Network Cookies: a mapping abstraction

Free Lane

25

Cookie Descriptors : ID + Key + Metadata Cookies: Unique, signed, use-once

slide-26
SLIDE 26

Cookies & Cookie Descriptors

  • Separation of data plane and control plane
  • Descriptors : Authentication + Service Definition
  • Cookies : Generate and Match locally at dataplane
  • Get cookie descriptors through an out-of-band mechanism
  • Insert cookies through an “agent” (browser, OS, application)
  • Where to insert a cookie?
  • Anywhere we can put a few extra bits
  • HTTP header
  • TLS handshake
  • IPv6 extension header
  • …

26

slide-27
SLIDE 27

Descriptor DB DB Cookie Descriptor Cookie Server

  • 4. Match cookie, and enforces the service

Cookie

  • 3. Generate and add

cookie to boosted flows

  • 2. Sync descriptors with AP

27

slide-28
SLIDE 28

Example Preferences

Background Lane Everything goes to the best effort lane, apart from…

  • 1. Dropbox and software updates à background lane
  • 2. Google Hangouts and living room TV à fast lane

28

slide-29
SLIDE 29

“I want traffic X to get service Y!”

Internet

  • 1. I want Netflix to be faster
  • 2. I want Spotify zero-rated
  • 3. I want low-latency Skype for work

29

  • 3. Configure the network
  • 2. Communicate it to

the network

slide-30
SLIDE 30

Putting it all together…

Free Lane

  • 1. Discover Cookie service and acquire cookie descriptors
  • 2. Generate unique, use-once, signed cookies
  • 3. Append them to desired traffic (HTTP header, TLS extension, TCP, …)
  • 4. Match in network

30

slide-31
SLIDE 31

Network Cookies Properties

31

slide-32
SLIDE 32

Network Cookies Properties

Mapping abstraction ✔ Low transaction cost à Inclusive ✔ Composable/Delegetable ✔

Simple & Expressive

32

slide-33
SLIDE 33

Network Cookies Properties

Mapping abstraction ✔ Low transaction cost à Inclusive ✔ Composable/Delegetable ✔ Independent of packet header, payload, path ✔ High Accuracy ✔ Multiple transport mechanisms ✔ Only client-network support required ✔

Simple & Expressive Deployable

33

slide-34
SLIDE 34

Network Cookies Properties

Mapping abstraction ✔ Low transaction cost à Inclusive ✔ Composable/Delegetable ✔ Independent of packet header, payload, path ✔ High Accuracy ✔ Multiple transport mechanisms ✔ Only client-network support required ✔ Built-in-Authentication ✔ Protected from replay, spoofing ✔ Transparent, Revocable, Auditable ✔

Simple & Expressive Deployable

34

Respect trust

slide-35
SLIDE 35

Outline

  • Why user preferences matter
  • Expressing user preferences to the network
  • Conclusions

35

slide-36
SLIDE 36

Conclusions & Next Steps

① A User-Driven approach is practical and beneficial

  • Evidence from real users (Boost, Zero-Rating)

① How we communicate user preferences is important!

  • Network cookies one way to do it

① Next Steps

  • Trials
  • Standardization
  • Net Neutrality Regulation

36

slide-37
SLIDE 37

37

yiannisy@stanford.edu

slide-38
SLIDE 38

38

Thanks!

yiannisy@stanford.edu

slide-39
SLIDE 39

39

BACKUP SLIDES

slide-40
SLIDE 40

What music do users want to zero rate?

slide-41
SLIDE 41

What music do users want to zero rate?

66% of user preferences not in Music Freedom (November 2015)

slide-42
SLIDE 42

Cookie à Network Configuration

Cookie Matching Configuration (forward+reverse flow) Cookie Matching Configuration (forward+reverse flow)

42

slide-43
SLIDE 43

Cookie à Network Configuration

Cookie Matching Configuration

43

slide-44
SLIDE 44

Cookie à Network Configuration

Cookie Matching Configuration reverse flow

44

slide-45
SLIDE 45

Out-of-band (or SDN*)

Internet

Flow tuple changes in the network

* SDN à Software-Defined Networking

45

Boost 192.168.1.10:4568 ßà54.204.2.219:80

cnn.com skai.gr 20 40 60 80 100 packets boosted (%)

matched false

OOB doesn’t work with middleboxes (e.g. NAT)

slide-46
SLIDE 46

What does a cookie look-like?

ID MAC UUID TIMESTAMP

Generation

Best effort Lookup ID? Valid MAC? Valid tstamp Uniqu e UUID Append UUID Apply Service

Matching

46

ID Key Descriptor

Low overhead, cannot be replayed or spoofed

slide-47
SLIDE 47

Where do I get a cookie descriptor?

  • Well-known server
  • Discovery protocols
  • Given by content provider

Out-of-band + authentication primitives

47

slide-48
SLIDE 48

Scalability

48

64 256 512 1024 1500 packet size (bytes) 2 4 6 8 10 throughput (Gbps)

10 pkts/flow 50 pkts/flow 100 pkts/flow