Content Peering William B. Norton Co-Founder and Chief Technical Liaison Content Peering Forum September 18, 2002 © Equinix 2002
Jan 11, 2016
Content Peering
William B. Norton
Co-Founder and Chief Technical Liaison
Content Peering ForumSeptember 18, 2002
© Equinix 2002
Internet Researcher
• White Papers: Focused studies on Peering
1. Identify Relevant Internet Operations Topic
2. Speak with prominent Peering Coordinators
3. Write/Evolve Draft White Paper
4. Walk Peering Coordinators through paper
5. Goto Step 3
Most widely read WP: “Internet Service Providers and Peering”
Peering Research• Two-Year Study w/~200 Peering Coordinators
– What is Peering?– How do you determine who to peer with?– When does Peering make sense?– When does Peering NOT make sense?– What is the Process of Peering?– How is Peering implemented?
Goal: Document Internet Operations Practices I will share with you today The Motives, The Methods,
The Mindset of the Peering Coordinator
Summary Findings of Peering Research….
No Internet Operations DocumentationTo make things more difficult, subtle language differences!
Peering is a game of relationships
Level Setting Talk…
• Content Peering is…– Content-Heavy Companies leveraging Simple
Internet Peering, Peering as ISPs do.– A Routing Optimization (direct vs. through an
Intermediary)
• Not…– Content Peering Alliance
(see http://www.content-peering.org/peering.html)
– Or any end-to-end Content Alliance
To this end I will share…
Agenda
1. Definitions• Transit & Peering Terminology and Financial Cost Models
2. Introduction to the Peering Breakeven Analysis• Specific Examples (San Jose Peering) leading to Generalizations
3. Peering Process• 3 stages of Peering
4. Top 5 Reasons Not to Peer5. Peering Resources Available to you
Peering Makes Sense for Large Scale Content Players
WP: “Internet Service Providers and Peering”
Definitions• Gives us tools (lexicon) to facilitate discussion• Market Confusion over misuse of terms:
• Terminology critical to have meaningful discussion on Peering
TransitPeeringTransportTier 1 ISPEffective Peering Bandwidth
Defs: ISPs
Understand that…
1. Def: The Internet is a network of networks
2. Def: Internet Service Providers sell access to the Internet
3. Internet Service Providers must themselves connect to the Internet.
• How do they connect to the rest of the Internet?
Definition of Transit
1) ISP A buys Transit Service
Upstream Transit ProviderISP A
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit
Providers
Def: Transit is the business relationship whereby one ISP
announces (usually sells) reachability to the *entire* Internet to a customer.
•$$$ ? Typically usage-based•Pricing ’01: $100-$1200/Mbps•Volume based on 95th Percentile measure•Transit is Simple, Convenient:• Upstream handles the delivery of packets to the Internet by some meansUsage: “I’m purchasing transit from Level 3.”
3) Traffic Flows
INTERNET
NETWORKS
2) ReachabilityAnnouncement
Cost of Transit Traffic Exchange
2001 Pricing Sampling Range: $100-$1200/MbpsMy Financial Models used $388/MbpsSource: 2002 Survey Range: $50-$1200/Mbps 95th Percentile
Blended Avg: ~$200/Mbps wholesale Transit
Isn’t Transit Good Enough?
Transit Cost Model
$0
$50
$100
$150
$200
$250
$300
$350
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43
# Mbps Exchanged
$/M
bp
s
Transit Cost per Mbps
Mbps $/Mbps1-3Mbps $3004-10Mbps $22010-30Mbps $20530-50Mbps $18050-75Mbps $14675-100Mbps $129100-300Mbps $112300-1000Mbps $105
Transit Costs
Traffic Growth
Content Heavy ISP
Access Heavy ISP
UPSTREAMS
V
IGMBT
Streaming Media
56k 384k 1.5m
$$$
$$$
Yahoo! Broadcast: 100,000+ concurrent unicast streams15 million streaming hrs/mo, 1300 Live events/day
Access
AOL+ DSL: 1,000/dayRoadrunner Cable Modems: 1M subscribersKaZaa!!!! Two main motivations for Peering…
Streaming:BroadcastTelephonyVideoMultimediaInteractiveGamingSPOT Events
Seek transportInterconnection $
x
Why Peer? 2 Motivations for Peering1.Financial: Reduce load on expensive Transit
serviceTraffic src/destMeasure vs IntuitUsage-based Billing
2.Engineering: Lower latency
1st Stage of Peering:Top 10 destination ISP list
Transit$$$
Transit$$$
ISP A
ISP B
Transit ISP
Def: Peering…
Definition of PeeringDef: Peering is the business relationship whereby ISPs reciprocally announce reachability to each others’ transit customers.
Peering
USNet EastNetWestNet
Peering
TransitRoutingTables
•Peering is *not* a transitive relationship•Peering *does not* provide access to the entire Internet
Cost of Peering: Def. Transport+Port
Usage: “I buy transit from UUNet and Peer with EastNet, WestNet.”
The Costs of Peering
)__(
__
___
____
:
_
CostsEquipmentRouterR
FeesRackMonthlyr
FeesPortIXMonthlyp
IXIntoCostsTransportMonthlyt
where
RrptCostsPeering
Observation: Transport Prices have dropped like a rock. Observation: New Router prices have dropped like a rock.Observation: Used Router Market is also very healthy (cheap).
Graphically…
Transit ISP
ISP A
ISP B
Transit $$$
Transit $$$
1) Transport into Exchange
2) Rack Space atExchange PointFor Router
3) Router at IX
4) Ethernet Switch Port (/Private Cross Connect)
R
RX
Peering $Ethernet-Based Peering Model
When does Peering make sense? (SJ Mkt)Flat Monthly Fees vs. Metered Monthly Fee
San Jose Market Prices for Peering
Transit ISP
ISP A
ISP B
Transit $$$
Transit $$$
1) Transport into ExchangeOC3@$2500/mo
2) Rack Space atExchange PointFor Router½ rack $2500/mo
3) Switch Port on Public Peering FabricGigE
R
RX
Peering $
Total Cost of Peering$7000/month
Which is more cost effective?Peering or Transit? Allocate…
4) Cisco 7500 $2000/mo
Cost of Traffic Exchange Peering vs. solely Transit
Transit ISP
ISP A
ISP B
R
RX
Peering $
Cost of Peering: $7000/month
1 Mbps
Unit Cost of Traffic ExchangeIn Peering Relationship:
$7000=$7000 per Mbps1 Mbps
1 Mbps
TRANSIT~ $200/Mbps
Cost of Traffic Exchange Peering vs. solely Transit
Transit ISP
ISP A
ISP B
R
RX
Peering $
Cost of Peering: $7000/month
2 Mbps
Unit Cost of Traffic ExchangeIn Peering Relationship:
$7000=$3500 per Mbps2 Mbps
2 Mbps
TRANSIT~ $200/Mbps
Cost of Traffic Exchange Peering vs. solely Transit
Transit ISP
ISP A
ISP B
R
RX
Peering $
Cost of Peering: $7000/month
7 Mbps
Unit Cost of Traffic ExchangeIn Peering Relationship:
$7000=$1000 per Mbps7 Mbps
7 Mbps
TRANSIT~ $200/Mbps
Cost of Traffic Exchange Peering vs. solely Transit
Transit ISP
ISP A
ISP B
R
RX
Peering $
Cost of Peering: $7000/month
14 Mbps
Unit Cost of Traffic ExchangeIn Peering Relationship:
$4000=$500 per Mbps14 Mbps
14 Mbps
TRANSIT~ $200/Mbps
Cost of Traffic Exchange Peering vs. solely Transit
Transit ISP
ISP A
ISP B
R
RX
Peering $
Cost of Peering: $7000/month
70 Mbps
Unit Cost of Traffic ExchangeIn Peering Relationship:
$7000=$100 per Mbps70 Mbps
70 Mbps
TRANSIT~ $200/Mbps
OC-3 Peering vs. Transit in San JoseOC-3 Peering vs. Transit
$0
$100
$200
$300
$400
$500
$600
$700
$800
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
# Mbps Exchanged
$/M
bp
s
Peering
Transit
Mbps $/Mbps1-3Mbps $3004-10Mbps $22010-30Mbps $20530-50Mbps $18050-75Mbps $14675-100Mbps $129100-300Mbps $112300-1000Mbps $105
Transit Costs
Mbps Peering10 $70020 $35030 $23340 $17550 $14060 $11770 $10080 $8890 $78
100 $70110 $64120 $58130 $54140 $50150 $47
Generalized…
OC-12 Peering vs. Transit in San JoseMbps $/Mbps1-3Mbps $3004-10Mbps $22010-30Mbps $20530-50Mbps $18050-75Mbps $14675-100Mbps $129100-300Mbps $112300-1000Mbps $105
Transit Costs
Generalized…
OC-12 Peering vs. Transit
$0
$100
$200
$300
$400
$500
$600
$700
$800
$900
$1,000
10 30 50 70 90 110
130
150
170
190
210
230
250
270
290
310
330
350
370
390
410
430
450
470
490
510
530
550
570
590
610
#Mbps Exchanged
$/M
bp
s
Peering
Transit
$15/Mbps
Mbps Peering50 $190
100 $95150 $63200 $48250 $38300 $32350 $27400 $24450 $21500 $19550 $17
Peering Breakeven Effective Peering BandwidthDS-3 12 42.3OC-3 35 145.7OC-12 70 584.68
Effective Peering Range
Cos
t of
Tra
ffic
Exc
hang
e in
$/M
bps
Amount of Traffic Exchanged (in Mbps)
Transit Price per Mbps
Peering Price per Mbps
Peering Breakeven Point(Peering=Transit)
EffectivePeering Bandwidth (in Mbps)
Min Cost ofTraffic Exchange(in $/Mbps)
Effective Peering Range
Transit Cheaper-Peering Cheaper
Peering Analysis Graph (axis)
Cos
t of
Tra
ffic
Exc
hang
e in
$/M
bps
Amount of Traffic Exchanged (in Mbps)
Transit Price per Mbps
Peering Price per Mbps
Peering Breakeven Point(Peering=Transit)
EffectivePeering Bandwidth (in Mbps)
Min Cost ofTraffic Exchange(in $/Mbps)
Effective Peering Range
Transit Cheaper-Peering Cheaper
Peering Analysis Graph (EPB)
Definition: The Effective Peering Bandwidth is the maximum bandwidth available for peering, defined as the minimum of the available transport bandwidth and the usable bandwidth on the shared peering fabric.
Cos
t of
Tra
ffic
Exc
hang
e in
$/M
bps
Amount of Traffic Exchanged (in Mbps)
Transit Price per Mbps
Peering Price per Mbps
Peering Breakeven Point(Peering=Transit)
EffectivePeering Bandwidth (in Mbps)
Min Cost ofTraffic Exchange(in $/Mbps)
Effective Peering Range
Transit Cheaper-Peering Cheaper
Peering Analysis Graph (minCost)
Definition: The Minimum Cost of Traffic Exchange is the unit cost of traffic exchange when the Effective Peering Bandwidth is fully utilized.
Cos
t of
Tra
ffic
Exc
hang
e in
$/M
bps
Amount of Traffic Exchanged (in Mbps)
Transit Price per Mbps
Peering Price per Mbps
Peering Breakeven Point(Peering=Transit)
EffectivePeering Bandwidth (in Mbps)
Min Cost ofTraffic Exchange(in $/Mbps)
Effective Peering Range
Transit Cheaper-Peering Cheaper
Peering Analysis Graph (EPR)Definition: The Effective Peering Range (EPR) is the range in which peering at an Internet Exchange makes sense (financially), measured as the range between the Peering Breakeven Point and the Effective Peering Bandwidth.
When does Peering Make Sense?
Peering Breakeven Effective Peering BandwidthDS-3 12 42.3OC-3 35 145.7OC-12 70 584.68
Effective Peering Range
The 3 Stages of Peering
The 3 Stages of Peering
Interviews with 200 ISP Peering Coordinators revealed…
3 General Phases of Peering:
1) Identification of Potential Peer – the who
2) Initial Contact and Qualification – the why
3) Implementation Discussions – the how
I. Phase 1: Identification of Peer: Traffic Engineering Data Collection and Analysis
Motivations:• Reduce load on expensive Transit service• Traffic src/dest• Measure or Intuit
• Usage-based Billing• 2nd Goal: Lower latency• ResultTop 10 list (see next page)• Part of larger business deal
Seekinterconnection
Transit$$
Transit$$
ISP A
ISP B
Transit ISP
Sample Top 10 Destination ListA S N u m b e r M b p s D e s t i n a t i o n I S P C o n t a c t
6 1 7 2 2 4 . 3 5 H O M E - N E T - 1 [ H O M E - N O C - A R I N ]
7 0 1 8 . 9 0 A L T E R N E T - A S [ I E 8 - A R I N ]
1 6 6 8 8 . 1 4 A O L - P R I M E H O S T [ A O L - N O C - A R I N ]
4 7 6 6 7 . 0 8 A P N I C - A S - B L O C K [ S A 9 0 - A R I N ]
3 3 2 0 5 . 1 2 R I P E - A S N B L O C K 4 [ R I P E - N C C - A R I N ]
5 7 7 4 . 2 4 B A C O M [ E Q - A R I N ]
6 3 2 7 3 . 9 0 S H A W F I B E R [ I A S - A R I N ]
1 3 . 8 9 B B N P L A N E T [ C S 1 5 - A R I N ]
7 0 1 8 3 . 6 6 A T T - I N T E R N E T 4 [ J B 3 3 1 0 - A R I N ]
9 3 1 8 3 . 1 3 A P N I C - A S - 3 - B L O C K [ S A 9 0 - A R I N ]
5 7 6 9 2 . 6 7 V I D E O T R O N [ N A V 1 - A R I N ]
6 8 3 0 2 . 3 0 H C S N E T - A S N B L K [ M D 2 0 5 - A R I N ]
9 2 7 7 2 . 2 2 A P N I C - A S - 3 - B L O C K [ S A 9 0 - A R I N ]
1 0 9 9 4 2 . 0 8 T A M P A 2 - T W C - 5 [ J D 6 - A R I N ]
1 2 3 9 2 . 0 5 S p r i n t L i n k [ S P R I N T - N O C - A R I N ]
I n t e r n e t S e r v i c e P r o v i d e r A
Phase 1: Identification of Potential
Peer
Part of BroadBusiness
Relationship?
Dom inant TrafficFlow?
TraversingExpensive Transit
C ircuit?
Yes
W ill Peering have apositive affect on m y
network?
Yes
Large new custom erim pact?
Yes
Yes
Proceedto Phase
2:Contact
Peer
Yes
II. Phase 2: Contact & Qualification, Initial Peering
DiscussionHow to make contact with potential peer ISP?
1. E-mail person or peering@<ispdomain>.net
2. Exchange point participant list
3. Tech-c/admin-c from DNS/ASN registries
4. Engineering Forums NANOG, IETF, RIPE, etc.
5. Trade shows: speakers and booth staff
6. Target ISP sales force
7. Target ISP NOC
II. Phase 2: Contact & Qualification, Initial Peering
DiscussionOnce contact is made…1. Sometimes Mutual NDA2. Exchange BiLateral Peering Agreement (BLPA)3. Traffic Data justification shared
One basis: Peering iff PeeringCost < TransitSavings?
4. Requirements Exchange(e.g. Must be at n Public Peering Points, xMbps, private
peering migration strategy, etc.)
Either Party may walk away….. If still interested, implementation discussion…
Phase 2: Contact and Qualification
InitialContact
LargerBusiness
Transaction
peering@ or
personalcontact
ExchangePoint
Contact list
tech-c oradm in-cin DNS/
ASNRegistry
OperationsForum
TradeShows
SalesForce
Finding the Right Contact
SignNDA,see
polic ies
Sharetrafficdata,BLPA
Do both parties findm otivation to continuepeering discussion?
Close discussion
Proceed toPhase 3:
Im plem entation
Discussion
YesNo
III. Phase 3: Implementation Discussions
How to interconnect?
Direct Circuit-based Interconnection
VS.
Exchange-Based Interconnection
Cost Comparison at n=5
costDCfn()=(n-1)*C/2 C=OC-3 @ $2,500n=5costDC=(4)*$2500/2costDC=$5,000/mo
costExchfn()=BDC+(n-1)*x/2+Racks BDC=OC-12 @ $5,000n=5, 1 Rack@$1000costExch=$5,000+(4)(250/2)+$1000costExch=$6,500/mo
More expensive to use Exchange-BasedInterconnection Strategy at n=5. N>5?
SSS
GGG
UUU
A A A
S
G
U
A
CCCC
S
G
U
A
C
OC-12
OC-12
OC-12
OC-12
OC-12
Exchange-based vs. Direct Circuit Interconnection
Cost Comparison of Interconnection Strategies
$-
$50,000
$100,000
$150,000
$200,000
$250,000
$300,000
$350,000
$400,000
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
# of participants
Mo
nth
ly C
ost
of
Inte
rco
nn
ecti
on
Direct Circuits Model
MUX Big Pipes Model
Dark Fiber Model
9 Exchange Selection Criteria
1. Telecommunications Access Issues2. Deployment Issues (getting in & up)3. ISP Current Presences (there yet?)4. Operations Issues (restrictions?)5. Business Issues (neutrality/alignment)6. Cost Issues ($$)7. Credibility Issue (backing,attraction) 8. Exchange Population (side effect)9. Existing vs. Emerging Exchange?
Value of the Internet ExchangeVExchange
NParticipants
CostOf Coming In(Circuits+Routers+StaffTime)
VCapacity
The Exchange Startup Hump
First Carrier(s)First ISP(s)First CP(s) Critical Mass Point (Vexchange=CostExchange)
Lar
ge F
acili
ty S
calin
g
C o nte ntPro vid e r
C o nte ntPro vid e r
C o nte ntPro vid e r
ISP ISPISP
C a rrie r C a rrie r
C irc uitDe m a nd sTric kle Do wn
C =num b e r o f c o nte nt p ro vid e rs
I=n u m b e r o f I S P s
N =n u m b e r o fc a r r ie r s
Top 5 Reasons NOT to Peer
Top 5 Reasons not to Peer
1) Already get Traffic for “free” (through existing peering relationships)
Transit$$$
Yahoo!
Transit ISPEXODUS
Peering $
AOL
Top 5 Reasons not to Peer
2) Not True Peers• Traffic inequity
• Scale inequity• Not even investments in infrastructure• Form: “I don’t want to haul your traffic
around the globe”
Large Global Network Provider
SmallRegional
Player
Huge investment in Int’s circuits,100’s of routers and colo sites,Staff installs, peering negotiations, Millions of customers, etc.
Top 5 Reasons Not to Peer
3) Lack of Technical Competence
Troubleshooting network problems takes longer when the other ISP NOC and engineers lack the technical expertise during an outage…
Top 5 Reasons Not to Peer
4) Transit Sales Preferred
• We rather sell you transit…“Let me introduce you to our sales guys”
Top 5 Reasons Not to Peer5) BGP is Tough
“BGP? No ExpertiseNo measurements No Justification to hire expertsBGP?”
TransitService
ISP A
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit
Providers
INTERNET
NETWORKS
TransitService
ISP A
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit
Providers
INTERNET
NETWORKS
TransitService
ISP A
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit
Providers
INTERNET
NETWORKS
TransitService
ISP A
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit Provider
Upstream Transit
Providers
INTERNET
NETWORKS
Primary
BackupTransit
Seek transportInterconnection
$
x
Transit$$$
ISP B
Transit$$$
ISP A
Transit ISP
Transit$$$
ISP B
Transit$$$
ISP A
Transit ISP
Transit$$$
ISP BTransit$$$
ISP B
Transit$$$
ISP A
Transit$$$
ISP A
Transit ISP
PrimaryBackupTransit
Conceptual Hurdle Conceptual Hurdle ComplexSimple
6: personality
Top 5 Reasons Not To Peer
5+ Personality Clashes:
They don’t understand each other
and they didn’t like the interaction
Resources Available to Peering Coordinators
Resources Available for Peering Coordinators
• Gigabit Peering Forums
• Other White Papers document Peering Practices
• Peering Contact Database
Gigabit Peering Forums
Other White Papers
“Interconnection Strategies for ISPs”“Internet Service Providers and Peering”“A Business Case for Peering”“The Art of Peering: The Peering Playbook”“Do ATM-based Internet Exchange Points make
sense anymore?”“The Peering Simulation Game”
Freely available from the author: [email protected]
Peering Contact Database
For Peering Coordinators OnlyToss in your Business Card &Receive a copy of everyone’s Business CardsEvery 6 weeks (or so)
Managed as a community service.E-mail to [email protected](Or give me your business card)
Conclusions
• Language of the Peering Coordinator– Transit, Peering, Transport
• ISP Peering makes sense if you can offload x Mbps of traffic to peers…
• The Peering Process includes 3 Phases:– Identification of Peer– Contact and Qualification– Implementation of Peering
• These represent the Baseline Understandings of the Peering Coordinator
Acknowledgements?
Acknowledgements
For this white paper I’d like to thank a few folks in particular for their review, insights, and comments on this paper:
Dorian Kim (NTT/Verio), Ingrid Erkman (ICG), Dave McGaugh (ELI), Eric T. Bell (Time Warner Telecom), Chris Parker (StarNet), Brokaw Price (Yahoo!),Lane Patterson (Equinix), Jay Adelson (Equinix), Morgan Snyder (Equinix),John Hardie (Equinix), David Diaz (BellSouth), Joe Wood (Accretive Networks), Robert Seastrom (inter.net), Kevin Epperson (Level3), Petri Helenius (FICIX), Scott Sheppard (BellSouth), Ralph Doncaster (iStop.com), Leo Bicknell (ufp.org), Paul Vixie (vix.com), Ian Somerton and Dave Wodelet (Shaw/BigPipe), Tony Hain (Cisco), Jeff S. Wheeler (five-elements.com), Cliff Hafen, Dory Liefer, Shannon Lake (Omnivergent), Nenad Trifunovic (WorldCom), Andre Gironda (eBay), Jeb Linton (EarthLink), Daniel Golding (SockEye), Peter Moyer (Juniper), and others that preferred no recognition for their contributions to this paper.
Top 5 Reasons Not To Peer…
Questions?
Do these same Motivations to Peer apply to Content Companies?
Are the Financial Motivations more compelling thanthe Performance Improvement Motivations?