Hee-SuKim

Changing Patterns of Internet Trafficg gand Net Neutrality

2011. 6. 24

Hee-Su Kim Ph DHee Su Kim, Ph.D.Vice President

Economics and Management Research LabKT CorporationKT Corporation

1. Changing Traffic Patterns a. Data Explosion

Data Explosion: N-screen, Cloud, OTT video streaming, P2P

Growth of Internet traffic per person: The world’s most - Korea (33G/mon) p p

Changes in Internet usage patterns: Web P2P Video

Increases in the share of video traffic: 37%(2010) 56%(2014) (Cisco, 2011)

Internet Traffic Volume (per person)Channing Traffic Patterns

* Source : Economist (June 2011)* Source : CiSCO (2011)

1

1. Changing Traffic Patterns b. Asymmetric Traffic Usage

A fraction of the heavy users takes up most of the traffic

Negative impacts on average Internet usersNegative impacts on average Internet users

Fixed Internet usage: the top 10% users’ share, 68%(’08) 80%(’10)

Mobile Internet usage: the top 1% users’ share, 27%(’10.3) 33%(’11.3)

Mobile: Share of Traffic by UsersFixed: Share of Traffic by Users

83

93

80

70

80

90

100

to u

sers 2500 

3000 Traffic(TB)

Traffic due to the Top 1%

Traffic of all users

33%

68

40

50

60

70

f Tr

affic

due

t

850 41000

1500 

2000 

33%

27%20

30

40

0 10 20 30 40 50 60 70 80 90 100

% o

f

2008 2010

68.6  66.9  82.6  142.6  153.3  173.9 220.0 

352.0 484.4 

621.3 719.3  696.0 

850.4 

0 

500 

1000 

27%

* Source: KT(2010)2

* Source: Atlas(2011)

0 10 20 30 40 50 60 70 80 90 100% of Users

10.3 4 5 6 7 8 9 10 11 12 11.1 2 3

1. Changing Traffic Patterns c. Peak-Time Extension

Peak-time lasts for an extended period due to Always-on n/w services:streaming, cloud, and so on.

Peak-time traffic is less effective to control (capacity expansion is more pressing) comparedwith the past (short duration of peak time).

Always on n/w Service Internet Traffic on Peak TimeAlways on n/w Service

SNS-Keep Alive Traffic ㆍFrequent keep-alive’i l f

Internet Traffic on Peak-Time

ㆍInternet traffic: Expect to change similar to the video traffic pattern

Video-StreamingㆍIncreases in

signals forSNS degradesmobile n/w

[ KT Traffic (2011) ]

TVInternet

ㆍIncreases invideo contents& streamingservices

Cloud-Netbook ㆍAlways-onInternetconnected PCenvironment 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5environment

3

* Olleh TV, Internet traffic

1. Changing Traffic Patternsd. Limitation of Mobile n/w

Recurrent traffic overload puts n/w redundancy at risk.

n/w utilization rate: Fixed 85% (Jan 2011), mobile 119% (Dec 2010)n/w utilization rate: Fixed 85% (Jan 2011), mobile 119% (Dec 2010)

An indefinite capacity expansion of the mobile network is impossible.

Fixed: building backbone vs. Mobile: limitation due to spectrum scarcity

Limitation of the Mobile n/w

g p y

Investment in the Fixed n/w

In EU, investment in the fixed n/w 1.4 times

more than 2009

- Similar patterns of expected investment in Korea

3G Traffic

The limit of mobile capacity(2,300TB)

TB/Month

The optimal level of mobile capacity(1,370TB)

* Source: KT (2011)

4

* Source: ATKEARNEY (2010)

2. Current Internet Pricing & Limitation a. Internet Access

Internet access fee: uniform/flat pricing for each bandwidth/capacity category

Established when finely tuned (usage based) price discrimination was not worthy of- Established when finely tuned (usage-based) price discrimination was not worthy of measurement efforts because nobody (in each category) did not push the n/w to the limit

No price signal linked with traffic generation Economic Inefficiency

- End-user: Flat-rate No additional payment despite high internet usage

- OSP: bandwidth based Ad revenues growing with traffic generated make the price signalfelt less acute

The Absence of Price Signal

End-Users

Broadband(KT) Lite(50M) Special(100M)

OSP

Biz Collocation (IDC) 7 prod (1G~10Mbps)

6

Biz Kornet Express 12 prod (1G~1Mbps)

b. Interconnection2. Current Internet Pricing & Limitation

Internet Interconnection: Settlement-free or limited settlement based on the criteria

crudely made in the early Internet days

- ISPs with a 2:1 difference in the network/traffic size can be settlement-free peers

A root cause of free riding (unfair sharing of network costs)

- Larger networks let incoming traffic reach (terminate) farther than smaller peers

without a compensation for the difference

ClassificationType of

interconnectionSettlement

Internet Interconnection & Free-Riding

A-B / C-DTransit

connectionBandwidth or

traffic

B-CPeering

connectionFree

PeeringISP (B) ISP (C)

Tier2

Tier1

$10 000/G connection

S it hi t h h t

OSP

OSP

Tier2

ISP (D)IDC

IDC

OSPIDC

ISP (A)

$10,000/G$5,000/G

If ISP B wants to charge more to an OSP in its IDC for terminating increased traffic, the OSP can switch to a cheaper ISP (C or D) that peers with B and

6

Switching to cheaper hostend-user

to a cheaper ISP (C or D) that peers with B and remain connected to ISP B.

c. Constraint on Pricing2. Current Internet Pricing & Limitation

Retail pricing constraint : intensive competition, social/political pressure for rate reduction

- The world’s lowest internet access fee due to severe facility-based competitionCoverage - KT: 100%, SKB: 88.8%, LG: 97.9% (KISDI, 2010)

% of Fiber connection - Korea(52%), top rank with Japan (55%)

- Repeated demands for rate reduction from NGOs and the government※ Volume-based pricing failed to be adopted due to consumer resistance in 1994

Internet fee - Korea: 49won/Mbps ≪ New York: 696 won, Paris: 157 won

2000.4 - Reduction: subscription fee 50,000, basiccharge 16,000, calling rate 22/10sec

Pressures by the government & the NGOs

2001 - Rate reduction & provision of free calls Promotion of rate reduction by the Ministry of Finance

※ Volume-based pricing failed to be adopted due to consumer resistance in 1994

Ministry of Finance

2008~2010 - Election pledge: Rate reduction by 20% Presidential Election

2011 - Reduction of basic fee: 1,000 won Rate reduction TF

Wholesale pricing: Crude & Rigid Difficult to reflect rapid changes

Tiering criteria - n/w size, number of subscribers, the rate of traffic exchange

Settlement - Free-peering among the same tiers

7

Settlement Free peering among the same tiers- Payment for transit (the lower tier the higher tier)

※ KCC Interconnection Order No.42 & 46

d. No Incentive to Invest2. Current Internet Pricing & Limitation

Huge growth imbalance in terms of revenue, operating profit and stockprices between network operators and other players in the Internet

Up until a few years ago, the growth of OSPs/CPs helped facilities-basedISPs to attract more subscribers. Now, with the saturated subscriber base, such win-win scenarios broke down. Network operators would not investpif the imbalance were to persist.

Internet Players’ Performance: Revenue, Operating Income, & Stock (’06~’10)

Revenue- The significant growth of device makers &

200%

250%

300% 1046%

≈

OSPs

Operating profit- The dramatic growth of OSPs & solution firms100%

150%

200%UI/Device

OSP

Solution

Content- No growth of n/w firms

Stock price- All players showed growth other than n/w -50%

0%

50%

Revenue Operating

profit

Stock

Content

Network

8

firms ※ Device: Samsung electronics, LG Electronics; OSP: NHN, Daum; Solution: KCP, Danal; Content: SBS, NCSoft; N/W: kt, LGU+, skt

profit

3. Suggestion a. Retail Pricing Reform

Give support to usage-based (and/or usage-cap) pricing

Efficient use of n/w, neutral treatment (no prioritization) of data packets

Gives n/w operators incentives to invest in next-generation Internet

Most Internet users are better off from reduced traffic congestionMost Internet users are better off from reduced traffic congestion

Examples of Data Plans on Fixed n/w Examples of Data Plans on Mobile

Telco Data plan

Abolishment of unlimited data plan

Country Company Data Plan

US AT&T $10/50GB in exceed of 150~250GB/mon (May 2011) AT&T

Abolishment of unlimited data plan (June 2010)

VerizonPlan for abolishing unlimited data plan (2011)

150~250GB/mon (May 2011)

Comcast $10/50GB in exceed of 250GB(Aug 2010)

Time Consideration of implementing l b d

VodafoneImplement of usage cap pricing-500MB/mon (Aug 2010)

Warner volume-based pricing

Canada Telus $2/GB in exceed of 125GB

Bell C d

$2/GB in exceed of 250GB

9

Canada

3. Suggestion b. Interconnection Pricing Reform

Give support to settlement-based peering (with even symmetric ISPs paying

each other for the traffic caused to the other party) Sends right price signals to p y) g p g

CPs/OSPs, Revives win-win Internet ecosystem, Neutral treatment of data packets,

Boosts investment incentives, Induces CPs to multi-home with several ISPs

OSPs are charged for traffic delivery based

Charging OSPs for n/w Costs

Traffic often flows very asymmetrically among

ISP d h l f

Why Paid Peering?

on the incremental costs that they caused

the hosting ISP to incur (i.e. paid peering

or transit charges)

ISPs under the current settlement-free

peering system when heavy-traffic generating

CPs single-home with one of the o t a s t c a ges)

- Telcos in EU consider charging CPs for

n/w usage

interconnected ISPs

- Netfilx takes up 25% of total traffic in North

America When it chose to single-home with

OSP ISPEnd UserTr

affic

Traf

fic

America. When it chose to single home with

Level 3, Comcast (Level 3’s peering ISP)

demanded extra charges for the huge

incremental traffic from Netflix to Comcast

Charges based on traffic Volume

c e e ta t a c o et to Co cast

- Google is ranked 3rd in overall global traffic

carried, generating more than Sprint (Tier 1) 10

3. Suggestion c. Network Management Rules

Rules for reasonable network management need to be established ASAP for

increased transparency for businesses as well as consumersp y

Declare usage-based pricing (retail and interconnection) as basically exempt

from net neutrality regulationy g

- The US (strictly applies net neutrality principles) officially allows for telcos to charge more

to heavy traffic users for their extreme internet usage

Net Neutrality & PricingGlobal Telcos’ n/w management

Telco n/w Management The FCC’s Open Internet Order (’10.12)

BT - Limits Internet speed for the heavy usersduring peak-time (5pm~12am)

SingTel- Provides 1/10 of standard traffic during

peak time while providing standard traffic

‘The Framework we adopt today does not prevent

broadband providers from asking subscribers who use SingTel peak-time while providing standard traffictwice during off-peak

NTT - Usage cap: 30GB upload per day - Limits the heavy users’ usage since 2008

p g

the network less to pay less, and subscribers who use

the network more to pay more.’

Comcast - Delayis traffic during n/w congestion (over 70% of the total n/w capacity)

11

4. Conclusion

A Win-Win situation for both telcos & OSPs in the past

- Telcos: increases in the value of networks due to various contents provided by OSPs

OSP b ib l- OSPs: access to subscribers at low cost

Telcos’ investments are not sustainable if the current situation will last- No win-win: saturation of subscriber base, traffic explosion without corresponding compensation

for telcos due to price signals not linked with traffic

Create (or do not regulate) a price-based (bit neutral) incentive mechanism for telcos to make sustainable investmentsfor telcos to make sustainable investments- Open Internet rules with no ex ante regulation of usage-based retail pricing and settlement-based

internet peering (fact or effect based ex post regulation preferred)

A New Incentive Mechanism in the Internet

- Telco: n/w values

Past: Win-WinData Explosion - Flat-rate

Now: Imbalance No Investment Incentives

- OSPs: access to subs - Settlement-free

New Paradigm Investment

12

g- Usage-based- Settlement-based

Investment Incentives& Win-Win