The Road to IPv6 Time to get going? A Tier1 Provider Case Study Yves Poppe
HEAnet National Networking Conference, Limerick, November 12th 2004
Dir. IP Strategy 1
Agenda
Any urgency ?
Case Study Teleglobe
2
Any urgency ?
Coming out of the telecom recession: An industry desperate for renewed revenue growth – Consensus : the next multibillion revenue opportunities imply IP based network convergence, multi-functional end-devices, always on, always p2p reachable, mobile and endowed with end to end security. – Continuing pressure on existing carrier business models with the advent of VoIP and new broadband wireless technologies – Visions of ubiquitous communications between billions of devices ranging from home networks to global sensor and RFID networks.
3
What does IPv6 bring to the table?
The only realistic path to make this perceived next multibillion dollar revenue wave a reality – Solves the address shortage and restores the peer 2 peer principle, prerequisite for always reachable and for fully distributed applications. – Provides scaleability for mobility – Autoconfig, neighbour discovery, plug and play: precondition for mobile ad-hoc networks and home usage. – Mandatory IPsec support;
ADSL, cable, 3G, Wi-Fi, Wi-Max provide the always on 4
Is the internet really in an impasse?
IPv4 addresses are effectively being rationed and will likely run out by 2008-2010
The shortage is hidden by the proliferation of NAT’s which allow re-use of addresses. – Worse than having an extension number behind a PBX – Like having a manually patched phone call nearly a century ago.
Telephony in 1920 needed permanent phone numbers and peer 2 peer communications Internet in 2004 needs permanent addresses and peer 2 peer communications 5
Limitations of a one way internet
The internet today : – 500 million people and devices who cannot reach each other directly. A couple of million major servers sitting in the network cloud mediate everything and are the biggest security risk.
Removing the shackles – Telephony growth was stunted by lack of scalability. Growth was only possible with direct dial, in other words peer 2 peer. • The first dial tone was in Germany in 1908. • It took until the 70’s to go from local DD to DDD to generalized IDDD
– Now Internet growth is stunted by the lack of scaleability. 6
An exhausting exhaustion debate Jan 1 : “flag day” IPv4 replaces Network Control Protocol (NCP) st
1983 1984
in Arpanet. Had 8 bit network and 24 bit host addresses. RFC 791: A, B, C, D, & E class system for address allocation. RFC 917 formalizes subnetting.
1989
RFC 1105 : BGP introduced as EGP did not scale anymore
1990
Prediction of the exhaustion of IPv4 Class B by 1994
1992
Prediction of the exhaustion of IPv4 addresses by 2005-2011
1993
RFC 1466: shortage of class B’s results in directive to assign blocks of Class C's instead of Class B's The number of entries in the "core" routing tables begin to grow exponentially and results in BGP4 and CIDR prefix addressing. Introduction of Network address translators (the evil NAT’s) breaks the end to end nature of the internet
1999 November 2004
Registries start to assign IPv6 addresses - IPv6 forum formed. The exhaustion debate still lingers on 7
For when the exhaustion of IPv4 addresses ? 2008 or 2020 ???
Tony
Hain offers an explanation (Beijing, april 2004) : RFC 3194 explains the allocation inefficiency of addresses. Depletion looks more like 2008. In fact, with the NAT smokescreen we could be reaching depletion as we speak.
Geoff Huston RIPE
sept2003 Amterdam:
This
sterile debate has missed the essence:
–ISP’s effectively ration IPv4 addresses and the shortage is hidden by the proliferation of NAT’s –Status quo is defendable but at a price: progressive crippling of the revenue growth engines.
Jim Bound IPv6 forum May 2003 Madrid
–The address dam will burst before too long under the pressure of p2p VoIP, e2e point to point secure connection, e2e VPN’s, mobile IP, RFID, grid applications, telemetry etc. 8
17 billion Traditional Networkable Devices!
Sun Microsystems estimates that including sensor and RFID networks the world could have a trillion communicating devices in a decade! 9 This should put the address debate to rest once and for all!
The perils of peer 2 peer
p2p will likely turn some existing businessmodels and revenue streams on their head – Carrier dilemma : always on connections (ADSL) and cellular now provide bulk of revenue growth but fixed line long distance still contributes a significant part.. • Ambivalence on VoIP: What happens to long distance revenues? • Reluctantly moving to network consolidation on IP backbone • Probable next battlefront: cell phone revenue
– Self preservation reflex: Stop the rising tide • NAT box providers • Customer control/lock-up • “managed transition”
King Canute (994?-1035)
10
The fear of crossing chasms
As presented by Latif Ladid, President IPv6 forum
11
The opportunities of peer 2 peer
Dissociation of network and services – Revenue opportunities higher in the value chain – New end to end services can easily be launched independently of the underlying network providers. i.e. VPN’s, p2p gaming, p2p VoIP, push and location based services, tracking and monitoring services etc
12
Agenda
Any urgency ?
Case Study Teleglobe
13
Teleglobe – Canada’s international Carrier; operates a worlwide voice and data network; major VoIP provider through ITXC acquisition. – Teleglobe provides the first NGI intercontinental connection in 1995 for the Brussels G7 summit. – A member of the Canarie Policy Board, Teleglobe promotes the experimentation of IPv6 and the 6bone/6TAP initiative – Teleglobe facilitates the world ’s first intercontinental native IPv6 connection ; becomes a founding member of the IPv6 forum. – Teleglobe presents its original IPv6 plans at the Telluride March 2000 IPv6 Forum – 2003: Teleglobe starts an IPv6 pilot and is the first Canadian carrier to announce commercial IPv6 introduction. 14
Canadian domestic carrier/ISP scene
Major players: – Telcos: Bell Canada, Telus, Sprint, Allstream(ex AT&T) – Cablecos: Videotron, Rogers, Shaw Cable
Some lab experiments and evaluation Lack of demand and compelling applications Still very busy with ADSL and cable growth Judge that they can take the step to IPv6 very quickly if/when needed; wait and see attitude Growing interest in IPv6 from Canadian Government and Department of Defence following the US lead. 15
Why bother with IPv6 in the first place?
IPv6 will lead to a rejuvenation of the internet with sizeable new service and new revenue opportunities. Position the company ahead of the pack. Differentiation in a commoditized IP Market place. Be ready when the growth phase really kicks in. Maintain a technological and QoS edge.
16
The major constraints
Great deal of uncertainty as to when the fast ascension part of the growth curve will be reached: 2006? 2008? Risky business case if major expense involved, especially under current tight economic conditions. Avoid any negative repercussions on SLA’s and QoS of existing operational IPv4 based network. IPv6 monitoring and debugging tools still evolving Perennial fear of crossing chasms.
17
Where does Teleglobe fit in the Internet World? DialDial-up/DSL Consumers
ISP Tier 2 or 3
Regional ISP
DialDial-up/DSL Business Users
Transit
DialDial-up/ DSL Business Users
Transit
Teleglobe Backbone
Business Customers Dedicated Access
Internet eXchange
Regional ISP
Business Customers Dedicated Access
Private Peering Public Peering
DialDial-up / DSL Consumers
ISP Tier 1
Global Tier 1 provider for national/regional carriers and ISPs 18
The internet core triangle
Caida
Skitter
19
A Global IP Network….
20
… Based on a global backbone
21
AS6453 Network Highlights
Industry Leading SLAs Very high performance for RTD, 0% pack loss, 100% reachability
Single AS Allows to implement routing policy globally
Global OC-48/192 Backbone Customer data stays on the network longer guaranteeing SLAs are met
Private Peering with all Tier-1 carriers, over 70 public peerings Guarantees customer data will reach its final destination
Cisco Powered Network MPLS enabled providing speed and global VPN connectivity
Dedicated to customer service ISO Certified Global Customer Service Center/ Global Network Management Center
45+ Internet POPs worldwide Conveniently located at Internet Exchanges or Carrier hotels 22
Most extensive Reach (93 Countries) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
Algeria Andorra Angola Argentina Armenia Aruba Australia Bahamas Bahrain Bangladesh Bolivia Botswana Brazil Brunei Burkina Faso Cameroon Canada China Colombia Cook Islands Cuba Denmark Dominican Republic Ecuador Egypt El Salvador Faroe Islands (DK) France French Polynesia Gabon Gambia
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
Germany Ghana Greece Guatemala Haiti Honduras Hong Kong Iceland India Iran Ireland Israel Italy Jamaica Japan Jordan Kenya Korea Lebanon Libya Malaysia Mali Mexico Micronesia Mozambique Nepal Netherlands Netherlands Antilles New Caledonia Nicaragua Nigeria
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
Norway Oman Pakistan Palau Panama Paraguay Philippines Poland Puerto Rico Qatar Romania Russia Rwanda Samoa Saudi Arabia Senegal South Africa Spain Sudan Syria Taiwan Tanzania Thailand Tonga (Kingdom of ) Trinidad and Tobago Turkey Uganda Ukraine United Kingdom United States of America Zimbabwe
23
How to Migrate a network this size to IPv6?
Alternatives considered: – – – – –
Going dual stack all the way Going partially dual stack 6PE access , MPLS through the core Tunnel brokers Just wait and see
24
Selected path toward IPv6
Major points of consideration: – Speed of transition to pervasive IPv6? major unknown – Need for a positive customer IPv6 experience. – Customer exposure to IPv6 : from nil to advanced.
Approach minimizing investment and operational risk: – Native peering with the IPv6 world using dedicated routers: connections to Equi6IX in Ashburn,Va and San-Jose,Ca, with MANLAN in New York and AMS-IX in Amsterdam. London, Madrid and Tokyo planned. – MPLS transport through the core – Customer access: • Teleglobe provided IPv6 over IPv4 Hexago tunnel broker using TSP (Tunnel Set-up Protocol) with AAA • Native IPv6 access to Teleglobe Cisco 6PE enabled dual-stack access routers. 25
IPv6 Support Timetable
Management approval: october 2003
Timeline – – – – – –
Beta service in 2004, full commercial availability in 2005 Q1 2004 : Customer access via tunnel broker Q1-Q4: native access via 6PE at selected sites Q2 2004: start of native peering with major partners Q2-Q3 2005: customer 6PE access at all sites. 2006: total dual stack depending on market conditions
Milestones met, some even ahead of shedule so far
26
1.4 Gig connectivity to the IPv6 world
Teleglobe Globeinternet
Customer networks
Equi6IX Ashburn
IPv6 World
Equi6IX San Jose Manlan NY AMS-IX Amsterdam
SFINX Paris
MPLS core
1)IPv4 only 6PE
2)partially or totally dual stack 3)networks with MPLS core
27
Carrier/ISP – Telegobe IPv6 Tunneling service Roaming user
IPv6 world
Firewall
Enterprise
SOHO
Home user
Carrier/ISP IPv4-only core
G L O B E I N T E R N E T
IPv4 core Migration Broker used in the core or aggregation points for major customers Carrier/ISP endusers are authenticated and assigned a fixed (stable) IPv6 prefix
IPv4 world
28
Rationale for Hexago tunnel broker
Inexpensive and proven solution for initial IPv6 experimentation and deployment – Based on the Canadian freenet6 tunneling service – Runs the TSP (Tunnel Set-up Protocol) which has AAA (Authorization, Authentication, Accounting), a key feature for mobile applications. – Has advanced NAT traversal features which allow for easier testing and early deployment of peer 2 peer applications. – Redeployable to the network periphery once IPv6 grows and networks go dual stack. – Usable for IPv6 in IPv6 VPN’s and for IPv4 In IPv6 tunneling once IPv4 becomes legacy. 29
Initial tunnel based IPv6 service
Teleglobe offers to – Provide its customers and the customer’s endusers exposure to initial IPv6 services assigning stable (non-dynamic) address space with automatic allocation of Teleglobe IPv6 addresses or addresses from the customer’s own IPv6 address space, allocated by the local RIR and offer auto configuration and plug and play connectivity
– Jointly try out and evaluate new end to end services made possible by globally reachable IPv6 addresses • • • •
Peer to peer applications including VoIP End to end encryption and security Unrestricted terminal mobility and plug and play. Test IPv6 QoS and VPN
– Judge the IPv6 demand and speed of transition
At minimal cost 30
Carrier/ISP - Teleglobe IPv6 native service
IPv4 World
Teleglobe Globeinternet Approx 45 locations worldwide 72 gig of peering
Carrier/ISP dual stack network
MPLS core
6PE
IPv6 World
Ashburn, San Jose, New-York, Paris, Amsterdam,
Enterprise
Dual stack router
or network with MPLS core
SOHO
Home 31
Rationale for 6PE MPLS operation
Cost effective solution in a Cisco Powered Network environment – No need to upgrade hardware nor software in the core network – Maintains benefits of the current MPLS features while appearing to provide a native IPv6 service – IPv6 forwarding is done by label switching, no tunnels in the core; IPv6 traffic is encapsulated using 2 levels of labels. – To become 6PE, the PE needs to be upgraded to dual-stack and configured to run MPLS on the interfaces connected to the core – The 6PE routers exchange reachability information between each other using MP BGP
32
Carrier/ISP -Teleglobe joint native IPv6 service
Teleglobe offers – – – –
Dual stack router to dual stack router connectivity Optional Hexago tunnel brokers to extend capillarity IPv6 + IPv4 access to the Carrier/ISP’s customer base. High quality access to the IPv6 world through high capacity connectivity and native peering.
33
IPv4
IPv6
i
IPv4 only universe
v6 IP
IPv4
P nI
v4
IPv6 +IPv4
v6 IP
in
v IP
4
IPv4
Dual Stack IPv4 + IPv6
IPv4 IP v6 i
IPv6 +IPv4
n
IP v4
The big bang IPv6 +IPv4
34
Early assessment
IPv6 awareness is high amongst providers, industry and Government, but still relatively low in the Corporate World. IPv6 is shaping up as a powerfull differentiator and is becoming a mandatory feature in an increasing number of Requests for Quotation. Most carriers/ISP’s want to somehow get on the IPv6 bandwagon but with minimum risk and exposure. Fifteen beta customers in 11 countries have been connected. Migration Brokers have been shipped and native connections (dual stack to dual stack) set-up.
The chasm is being crossed! 35
Crossing the chasm
Still relatively unclear : how wide is it? One certainty : it will be crossed Why? – If you’re positive: It is a huge opportunity – If you’re reluctant: No choice but to cross the gap: economic well-being and even survival might be at stake.
A growing number of companies are preparing to take the leap to early majority; others continue to procrastinate 36
Thank you for your attention
37