Understanding Remote Peering – The New Wave of Interconnection at the Core of the Internet.
Using real-world case studies, this free webinar explains remote peering and what it means to ISPs, content providers and the global Internet peering ecosystem. Learn from William B. Norton who has presented three popular USTelecom webinars on Internet peering.
Background
The Internet peering ecosystem is going through a historic and rapid paradigm shift.
The largest ISPs and content providers have always interconnected their networks at the core of the Internet using a technique called "Internet peering," the free and reciprocal exchange of access to each other's customers. In this way, networks of scale can exchange a large enough amount of traffic for free with one another to offset the cost of deployment (equipment, colocation, and transport to the colocation center). This justification is the basis for the business case for peering.
However, a recent trend -- called "remote peering" -- has emerged as a way to get these peering benefits but without the cost of additional equipment, transport, or colocation. The remote peering model is where a remote peering provider delivers transport to the customer router with Virtual Local Area Network (VLAN) extension(s) from the largest exchange points in the world. In this way, the customer gets all of the benefits of peering (performance, control over routing, direct relationships with the peer networks, etc.) without the large initial capital and operational costs.
This is not just a fringe or small change to peering - it is a fundamental shift in the Internet architecture. Remote peering is a new technique that helps make peering accessible to a much larger population. As a result of the cost shift, an increasing percentage of networks are peering across great distances. The peering paradigm of "peering keeps local traffic local" is no more.
During the free webinar you will hear case studies from the field where medium-sized content companies are able to enter the peering ecosystem and connect to multiple Internet Exchange Points over a single circuit. These companies have graciously allowed their cost numbers to be shared so the traditional peering model can be compared against the emerging remote peering model. Also, the webinar will highlight the strongest arguments on both sides of the debate over whether remote peering is good or bad for the global Internet peering ecosystem.
William B. Norton, Executive Director, DrPeering International and Author of the new 2014 Edition of “The Internet Peering Playbook: Connecting to the Core of the Internet” which includes a new chapter dedicated to remote peering.
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Understanding Remote Peering - Connecting to the Core of the Internet
1. Understanding Remote Peering
William B. Norton
Chief Strategy Officer, IIX
Executive Director, DrPeering International
wbn@iixPeering.net
wbn@DrPeering.net
US Telecom Webinar
Live from Silicon Valley
July 23, 2013 10AM PST
2. Meet the Presenter
• Started working on Internet (NSFNET) in
1988
• 1st “Chairman” of North American Network
Operator Group (NANOG) (1994-1998)
• 1998-2008 Co-Founder & Chief Technical
Liaison, Equinix Inc. (NSDQ: EQIX)
• 2008-Present - Executive Director, DrPeering
Int’l
• Two-day On-Site Peering Workshops (EU/Africa)
• The Internet Peering Playbook
• 2013 Chief Strategy Officer, International
Internet Exchange (IIX)
Remote Peering is an important topic…
3. Remote Peering
• Hot Topic
• 130 of the 630 at AMS-IX are remote peering
• 50% of new at DE-CIX are remote peering
• This is not a fringe peering technique anymore
Source: Job Witteman (AMS-IX), Andreas Sturm (DE-CIX) Pre-webinar seminar…
4. Pre-Webinar Survey…
• Interest in Remote Peering?
– Understand emerging Internet Operations trend
– Understand Peering
– Building a NG Network – see how R.P. applies
• Make sure to cover
– Remote Peering impact on costs & architecture
– Applicability to Data Centers, IXPs, ISPs, CPs, etc.
• Remote Peering Pain Point
– Deployment challenges
Agenda…
5. Agenda
1. Connecting to the Edge: Internet Transit
2. Connecting to the Core: Internet Peering
3. Connecting to the Core from afar: Remote Peering
– Remote Peering Use Cases
– Implications of Remote Peering
3 Internet Interconnection Techniques
Internet Transit…
7. Internet Transit Service Model
• 99.9% of all
• Announce
Reachability
• Metered
Service
• Simple
• “Internet
This Way”
7
95th percentile measurement
11. What is Internet Peering?
• Definition: Internet Peering is the business relationship whereby two
companies reciprocally provide access to each others’ customers.
11
12. Internet Peering3 Key Points
1. Peering is not a transitive relationship
2. Peering is not a perfect substitute
3. Peering is typically settlement free
12
13. The Top 5 Motivations to Peer
1. Lower Transit Costs
(#1 ISP Motivation to Peer)
2. Improve end user experience
(#1 Content Motivation)
3. Better control over routing-strategic
(Yahoo!, NetFlix 2008)
4. Usage based billing – make more money by peering
(AboveNet)
5. Sell more underlying transport capacity
(Telecom Italia)
13
Important Traffic is Peered
14. “Important Traffic is Peered”
-- Andreas Sturm, DE-CIX
Transit – your traffic is in same bucket
Peering – you control this traffic
Does Peering cost less?
15. The Cost of Peering
Transport into DE-CIX
$2000/mo local
$4000/mo nearby
$6000/mo far
Assumptions Far
Transport into IX: $6,000 per month
Colocation Fees: $1,000 per month
Peering Fees: $2,000 per month
Equipment Costs: $2,000 per month
Total Cost of Peering: $11,000 per month
15
Costs allocated across volume peered…Source: 2010 DE-CIX meeting, Frankfurt
16. Cost of Internet Peering
100 Mbps $110.00 per Mbps
200 Mbps $55.00 per Mbps
300 Mbps $36.67 per Mbps
400 Mbps $27.50 per Mbps
500 Mbps $22.00 per Mbps
600 Mbps $18.33 per Mbps
700 Mbps $15.71 per Mbps
800 Mbps $13.75 per Mbps
900 Mbps $12.22 per Mbps
1000 Mbps $11.00 per Mbps
1100 Mbps $10.00 per Mbps
1200 Mbps $9.17 per Mbps
1300 Mbps $8.46 per Mbps
1400 Mbps $7.86 per Mbps
1500 Mbps $7.33 per Mbps
1600 Mbps $6.88 per Mbps
1700 Mbps $6.47 per Mbps
1800 Mbps $6.11 per Mbps
Peering CostMbps Exchanged
Assumptions Far
Transport into IX: $6,000 per month
Colocation Fees: $1,000 per month
Peering Fees: $2,000 per month
Equipment Costs: $2,000 per month
Total Cost of Peering: $11,000 per month
Cost of Peering allocated across the amount
of traffic peered for free.
16
Generalized in Peering Vs. Transit Graph.
17. • Definition: The Peering
Break Even Point is the
point where the unit
cost of peering exactly
equals the unit price of
Internet Transit.
17
We will use this graph
to compare Transit,
Peering and
Remote Peering
Market Dynamics affecting these graphs…
18. Two Peering Market Dynamics
Peering Cost Drops
…but Transit Prices Drop Faster
18
These two dynamics present a modern peering brick wall…
21. Definitions – What is Remote Peering?
• Definition: Remote Peering is peering without
a physical presence required at the peering
point.
• Definition: A Remote Peering Provider is an
entity that sells access to exchange points
across their transport infrastructure.
21
Review the costs of the Traditional Peering Model…
23. Remote Peering
Point to Multipoint Remote Peering Example
Peering Fabrics
Extended as VLAN
To customer
No router
No colo
Remote Peering Service Model…
25. Remote Peering
How does it work?
Remote Peering Provider is
already installed at
the IXPs.
Waves
provisioned,
instant turn up.
Neutral RPP
no business clash
Peering Focus
Speeds IXP deploy
Little paperwork
One Contract
Peering Fabrics
Extended as VLAN
To customer
No router
No colo
Remote
Peering
Provider
Node
Case Studies and Use Cases
27. Traditional Peering
LinkedIn Case Study
(NANOG 48 Peering BOF)
Colocation Expense
Router CapEx
Dublin
London
Frankfurt
AmsterdamSource: LinkedIn Spain (not shown)
$6K/mo
$3.5K/mo
$275K/mo
28. Peering Fabrics
Extended as VLAN
To customer
No router
No colo
Summary…
Remote Peering
LinkedIn Case Study
(NANOG 48 Peering BOF)
$6K/mo
Dublin
London
Frankfurt
Amsterdam
Spain (not shown)
29. Traditional Peering vs. Remote Peering
Source: Zaid Ali Kahn, LinkedIn
For the price of transport alone, one can remotely peer.
Use Cases…
31. For Content Companies
• Important Traffic is peered
– Gold plated packets
– Remote Peering simplifies
– Remote Peering Delivers
• Lower people cost
• Strategic intent
• Performance
• “End-User Performance”
• A/B testing and “the blend”
Commodity Transit is cheap,
but some applications require
more control / better performance
Cable Companies…
32. For Cable Companies / ISPs
• Case: Canadian MSO
peers in U.S.
• But Remote Peering
in LA and Miami for
free Asia and South
American Internet
routes
Int’l Networks…
33. For Int’l Networks
• Remote peer into
cheap ecosystem
– Buy Transit
– Peer Away
• 98% Africa traffic
from US/Europe
NG Networks…
34. For NG Networking
• Blend Transit,
• Strategic Internet
Traffic is remotely
peered
• No capital costs
• Control over routing
ISPs peer more…
35. For Internet Service Providers,
Remote Peering justifies
building into more and
even smallish regional
IXPs.
More controls
over routing
No Capital costs
36. IMPLICATIONS FOR THE INTERNET
PEERING ECOSYSTEM
What Remote Peering Means to the Internet Peering Ecosystem
If there is time…
37. 1) For all Network Operators
Remote Peering
extends the life of peering
Effective
Peering Range
Peering makes sense across
a wider range of Mbps. Peer away 2-3Gbps and all costs are covered.
Breathe of fresh air into Peering
More peersmore IXP value…
38. 2) For IXPs,
Remote Peering increases
the value of IXPs
More new peers,
more traffic,
more routes
More Valuable IXPs
Colo->instant critical mass
39. 3) For Colocation
Centers
the Remote Peering
Provider enables
instant critical mass
ALL PEERING ROUTES
ARE REACHABLE!
Reachable peersRemotely
makes colocation center
immediately valuable
(Value > Cost)
Trials and transition
40. 4) For all, rapid turn-up
means Remote Peering
enables a fast deployment/transition
strategy:
1. Remote Peering
2. Full Network Deployment
Summary…
Summary Implications and Predictions
Remote Peering will accelerate existing colocation centers to critical mass,
allowing them to become peering points and
command a premium (~$2000/rack vs. $400/rack)
Many more colocation centers will then pop up, with massive power for high-density
Video server deployments with immediately with access to peering critical mass.
Maybe 1000 peering points across the U.S. to handle massive volumes of video
in the next few years.
41. Remote Peering Summary
• New but rapid adoption
• Connect to the Core from afar
• $0 Capital Cost
• Minimal deployment time
• IXP VLANs extended to the
customer router
42. Book Offer
• Send email to books@iixPeering.net
– Subject: webinar book
• Amazon.com: $83.31
• IIX Sponsored: $9.99
• Ch12: Remote Peering
– *FREE PDF* for review
– wbn@DrPeering.net
Editor's Notes
The Internet Transit service is shown in the diagram below provides access to the global Internet by:announcing the customer route across the Internet so any network on the Internet knows how to reach the customer network, andannouncing to the customer the information necessary to be able to send traffic to any destination in the Internet.In this mini ecosystem, we see the Cyan ISP purchasing transit from the Orange Transit Provider. The Orange ISP announces to the Cyan ISP reachability to the entire Internet (shown as many colored networks to the right of the Transit Providers). The Transit Providers propagate the Cyan route (shown as a cyan circle) across the Internet so that all networks know how to reach the Cyan ISP. With this reciprocal Internet Transit service, all Internet attachments know how to reach the Cyan ISP, and the Cyan ISP knows how to reach all Internet destinations.
Why did the 95th percentile come into existence? In the early Internet days, Internet traffic was charged on a circuit capacity basis. But if you didn’t use very much of this capacity, you were still paying as if you did. This made Internet Transit tough to sell so the usage-based (metered) model began. Initially some ISPs charged on average use, which ended up being skewed by the occasional burstiness associated with a spot event. To address this, one ISP adopted the 95th percentile measure that was primarily introduced to not overly punish a customer for the occasional spike in traffic volume, and still allow the ISP to bill based the load placed on its network. This approach seemed palatable and sold. The rest of the industry followed suit and 95-5 was born.
WestNet is an ISP with green customers, MidNet is an ISP with blue customers, and EastNet is an ISP with red customers.WestNet is in a Peering relationship with MidNet in which WestNet learns how to reach MidNet's blue customers, and MidNet reciprocally learns how to reach WestNet's green customers.EastNet is in a Peering relationship with MidNet in which EastNet learns how to reach MidNet's blue customers, and MidNet reciprocally learns how to reach EastNet's red customers.After these two peering sessions are established, the routing tables are in place as shown in the boxes beneath the ISP clouds. Since MidNet peers with both EastNet and WestNet, MidNet customers can reach both EastNet and WestNet customers.
The Peering Break Even point helps make the business case for peering objective and clear - if an ISP can peer more traffic than indicated by the Peering Break Even Point, then the ISP should prefer to peer instead of solely purchasing transit from an upstream ISP.
The Peering Break Even point helps make the business case for peering objective and clear - if an ISP can peer more traffic than indicated by the Peering Break Even Point, then the ISP should prefer to peer instead of solely purchasing transit from an upstream ISP.