CS 4700 CS 5700 Network Fundamentals Lecture 16

CS 4700 / CS 5700 Network Fundamentals Lecture 16: IXPs (The Underbelly of the Internet) Revised 3/23/2015

2 q q Outline Emerging Internet Trends Internet Exchange Points (IXPs)

The Internet as a Natural System 3 You’ve learned about the TCP/IP Internet � Simple abstraction: Unreliable datagram transmission � Various layers � Ancillary services (DNS) � Extra in-network support So what does the Internet look like?

What does the Internet look like? 4

What does the Internet look like? 5

Characterization challenges 6 Limited measurements and models can hint at it � Traceroute does not give us a complete view � Gao-Rexford (policy routing) doesn’t capture everything What is the Internet actually being used for? � Emergent properties impossible to predict from protocols � Requires measuring the network � Constant evolution makes it a moving target

How is the Internet used? 7

How is the Internet used? 8

Measuring the capital-I Internet* 9 Measuring the Internet is hard Significant previous work on � Router and AS-level topologies � Individual link / ISP traffic studies � Synthetic traffic demands But limited “ground-truth” on inter-domain traffic � Most commercial arrangements under NDA � Significant lack of uniform instrumentation *Mainly borrowed stolen from Labovitz 201

Conventional Wisdom (i. e. , lies) 10 Internet is a global scale end-to-end network � Packets transit (mostly) unmolested � Value of network is global addressability /reachability Broad distribution of traffic sources / sinks An Internet “core” exists � Dominated by a dozen global transit providers (tier 1) � Interconnecting content, consumer and regional providers

Traditional view 11

Does this still hold? 12 Emergence of ‘hyper giant’ services How much traffic do these services contribute? Hard to answer! � Reading: Labovitz 2010 tries to look at this.

How do we validate/improve this picture? 13 Measure from � 110+ ISPs / content providers � Including 3, 000 edge routers and 100, 000 interfaces � And an estimated ~25% all inter-domain traffic Do some other validation � Extrapolate estimates with fit from ground-truth data � Talk with operators

Where is traffic going? 14 Increasingly: Google and Comcast � Tier 1 still has large fraction, but large portion of it is to Google! � Why? As of 2009 Google is 6% of traffic Consolidation of traffic � Fewer ASes responsible for more of the traffic Over time Google begins delivering YT’s traffic

Why is this happening? 15

Transit is dead! Long live the eyeball! 16 Commoditization of IP and Hosting / CDN � � � Consolidation � � Bigger get bigger (economies of scale) e. g. , Google, Yahoo, MSFT acquisitions Success of bundling / Higher Value Services – Triple and quad play, etc. New economic models � � Drop of price of wholesale transit Drop of price of video / CDN Economics and scale drive enterprise to “cloud” Paid content (ESPN 3), paid peering, etc. Difficult to quantify due to NDA / commercial privacy Disintermediation � � Direct interconnection of content and consumer Driven by both cost and increasingly performance

New applications + ways to access them 17 Fixed vs. Mobile Usage

The shift from hierarchy to flat Verizon Money follows the arrows. $ Tier 1 ISPs (settlement free peering) AT&T $$$ Sprint $ $ Tier 2 ISPs Regional Access Provider $ Local Access Provider $ Autonomous systems (ASes) connect to each other based on business relationships. Tier 3 ISPs $ Local Access Provider $ Businesses/consumers

The shift from hierarchy to flat Verizon Tier 1 ISPs (settlement free peering) AT&T Sprint Tier 2 ISPs Regional Access Provider Regional Access Content. Provider provider no longer Local Access Provider doesn’t have to pay for consumer access to content! Tier 3 ISPs needs to pay for transit! More “eyeballs” less $$ Local Access Provider $ $ IXP Local Access Provider $ Businesses/consumers

A more accurate model? 20

How do ASes connect? 21 Point of Presence (Po. P) � Usually a room or a building (windowless) � One router from one AS is physically connected to the other � Often in big cities � Establishing a new connection at Po. Ps can be expensive Internet e. Xchange Points (IXP) � Facilities dedicated to providing presence and connectivity for large numbers of ASes � Many fewer IXPs than Po. Ps

IXPs Definition 22 Industry definition (according to Euro-IX) A physical network infrastructure operated by a single entity with the purpose to facilitate the exchange of Internet traffic between Autonomous Systems The number of Autonomous Systems connected should be at least three and there must be a clear and open policy for others to join. https: //www. euro-ix. net/what-is-an-ixp

IXPs worldwide 23 https: //prefix. pch. net/applications/ixpdir/

Inside an IXP 24 Connection fabric � Can provide illusion of all-to-all connectivity � Lots of routers and cables Also a route server � Collects and distributes routes from participants

Structure 25 IXPs offer connectivity to ASes enable peering

Inside an IXP 26

IXPs – Publicly available information 27

How much traffic is at IXPs? * 28 We don’t know for sure! � Seems to be a lot, though. � One estimate: 43% of exchanged bytes are not visible to us � Also 70% of peerings are invisible *Mainly borrowed stolen from Feldmann 20

Revised model 2012+ 29