IXP Design and Operational BCP
13 likes2,059 views
Internet Exchange Point design considerations include: 1) IXPs are layer 2 exchange points where ISPs interconnect their routers via an ethernet switch. 2) Factors like location, security, expandability, and neutral operation need to be addressed in the design. 3) Peering can be bilateral between individual ISPs or multilateral via a route server, with policies ranging from open to selective or closed.
IXP Design and Operational BCP
- 1. Internet Exchange Point Design Philip Smith <philip@apnic.net> APNIC 42 Colombo, Sri Lanka 28th September – 5th October 2016 1Last updated 27th September 2016
- 2. Presentation Slides p Will be available on n http://bgp4all.com/dokuwiki/conferences/ n And on the APNIC42 website p Feel free to ask questions any time
- 3. IXP Design p Background p Why set up an IXP? p Layer 2 Exchange Point p Design Considerations p Route Collectors & Servers p What can go wrong? 3
- 4. A bit of history Where did the IX concept come from? 4
- 5. A Bit of History… p NSFnet – one major backbone n US “National Science Foundation” funded n Connected academic & research institutions n Also connected “private company” networks, with acceptable use policy p AUP: No commercial activity n Three Network Access Points (NAPs): Chicago, New York, San Francisco p Private companies needed to interconnect their networks n Requirement to send “commercial traffic” n Could not cross NSFnet n Resulted in the early “commercial Internet Exchanges” 5
- 6. More History… p Early Internet Exchanges created in early 90s n CIX-West – west coast USA n MAE-East – east coast USA n D-GIX – Stockholm p End of the NSFnet in 1995: n Meant move towards commercial Internet n Private companies selling their bandwidth n ANS (operator of the late NSFnet) had to join IXes p Routing Arbiter project helped with coordination of routing exchange between providers n Traffic from ISP A needs to get to ISP B n The RADB is the remnant of that project 6
- 7. More History still… p The NAPs established late in NSFnet life were some of the original “exchange points” n NAP operators supported commercial activities as well n (Sprint: NY, PacBell: SF, Ameritech: Chicago, MFS: Vienna/VA) p The NAPs replaced by IXPs: n NAPs didn’t succeed (operated by ISPs), replaced by more neutral IXPs n E.g. Virginia NAP replaced by MAE-East (by MFS) p Mid 90s saw rapid Internet growth, with major providers connecting… 7
- 8. Even more History p D-GIX formed in Stockholm in 1992 n Three major ISPs interconnected n Latency reduction, performance gains n Local traffic stays local p LINX formed in London in 1994 n Five UK operators interconnected n Latency reduction, performance gains n Local traffic stays local p HKIX formed in Hong Kong in 1995 n Vibrant Internet community, many small operators n Latency, performance, and local traffic benefits p Also AMS-IX in Amsterdam in 1994 n Same reasons as others 8
- 9. Internet Exchange Point p What: n A neutral location where network operators freely interconnect their networks to exchange traffic p What is the physical IX: n An ethernet switch in a neutral location p How does it work: n IX Operator provides the switch and rack space n Network Operators bring routers, and interconnect them via the IX fabric p Very simple concept – any place where providers meet to exchange traffic 9
- 10. Internet Exchange Point p Layer 2 exchange point n Ethernet (100Gbps/10Gbps/1Gbps/100Mbps) n Older technologies used in the past included ATM, Frame Relay, SRP, FDDI and SMDS p Layer 3 exchange point n Has historical status now n Router based p Best known example was CIX-West p Router very quickly overwhelmed by the rapid growth of the Internet 10
- 11. Why an Internet Exchange Point? Saving money, improving QoS, Generating a local Internet economy 11
- 12. Internet Exchange Point Why peer? p Consider a region with one ISP n They provide internet connectivity to their customers n They have one or two international connections p Internet grows, another ISP sets up in competition n They provide internet connectivity to their customers n They have one or two international connections p How does traffic from customer of one ISP get to customer of the other ISP? n Via the international connections 12
- 13. Internet Exchange Point Why peer? p Yes, International Connections… n If satellite, RTT is around 550ms per hop n So local traffic takes over 1s round trip p International bandwidth n Costs significantly more than domestic bandwidth n Congested with local traffic n Wastes money, harms performance 13
- 14. Internet Exchange Point Why peer? p Solution: n Two competing ISPs peer with each other p Result: n Both save money n Local traffic stays local n Better network performance, better QoS,… n More international bandwidth for expensive international traffic n Everyone is happy 14
- 15. Internet Exchange Point Why peer? p A third ISP enters the equation n Becomes a significant player in the region n Local and international traffic goes over their international connections p They agree to peer with the two other ISPs n To save money n To keep local traffic local n To improve network performance, QoS,… 15
- 16. Internet Exchange Point Why peer? p Private peering means that the three ISPs have to buy circuits between each other n Works for three ISPs, but adding a fourth or a fifth means this does not scale p Solution: n Internet Exchange Point 16
- 17. Internet Exchange Point p Every participant has to buy just one whole circuit n From their premises to the IXP p Rather than N-1 half circuits to connect to the N-1 other ISPs n 5 ISPs have to buy 4 half circuits = 2 whole circuits → already twice the cost of the IXP connection 17
- 18. Internet Exchange Point p Solution n Every ISP participates in the IXP n Cost is minimal – one local circuit covers all domestic traffic n International circuits are used for just international traffic – and backing up domestic links in case the IXP fails p Result: n Local traffic stays local n QoS considerations for local traffic is not an issue n RTTs are typically sub 10ms n Customers enjoy the Internet experience n Local Internet economy grows rapidly 18
- 19. Who can join an IXP? p Requirements are very simple: any organisation which operates their own autonomous network, and has: n Their own address space n Their own AS number n Their own transit arrangements p This often includes: n Commercial ISPs n Academic & Research networks n Internet infrastructure operators (eg Root/ccTLDs) n Content Providers & Content Distribution Services n Broadcasters and media n Government Information networks 19
- 20. When an IXP is not beneficial p Legislation: When there is one legislated monopoly transit provider n With all other network operators legislated to be customers of this monopoly provider p Geography: When the local economy is so small that it cannot sustain more than one network operator n Very small nations (maybe less than 10000 population?) n Sparsely populated / remote areas 20
- 21. When an IXP is not permitted p This is still the situation in several countries around the world p Usually it is a Government operated “national telco” n ISP licence mandates connecting to “national telco” for Internet services p Implications: n Expensive domestic connectivity n Expensive international connectivity n Restricted and poor service offerings n No domestic Internet economy 21
- 22. Layer 2 Exchange The traditional IXP 22
- 23. IXP Design p Very simple concept: n Ethernet switch is the interconnection media p IXP is one LAN n Each ISP brings a router, connects it to the ethernet switch provided at the IXP n Each ISP peers with other participants at the IXP using BGP p Scaling this simple concept is the challenge for the larger IXPs 23
- 24. Layer 2 Exchange 24 ISP 1 ISP 2 ISP 3 IXP Management Network ISP 6 ISP 5 ISP 4 Ethernet Switch IXP Services: Root & TLD DNS, Routing Registry Looking Glass, etc
- 25. Layer 2 Exchange 25 ISP 1 ISP 2 ISP 3 IXP Management Network ISP 6 ISP 5 ISP 4 Ethernet Switches IXP Services: Root & TLD DNS, Routing Registry Looking Glass, etc
- 26. Layer 2 Exchange p Two switches for redundancy p ISPs use dual routers for redundancy or loadsharing p Offer services for the “common good” n Internet portals and search engines n DNS Root & TLDs, NTP servers n Routing Registry and Looking Glass 26
- 27. Layer 2 Exchange p Neutral location n Anyone can install fibre or other connectivity media to access the IXP p Without cost or regulations imposed by location p Secure location n Thorough security, like any other network data centre p Accessible location n Easy/convenient for all participants to access p Expandable location n IXPs result in Internet growth, and increasing space requirements 27
- 28. Layer 2 Exchange p Operation: n Requires neutral IXP management n “Consortium” p Representing all participants p “Management Board” etc p Funding: n All costs agreed and covered equally by IXP participants n Hosting location often contributes – the IXP brings them more business p Availability: n 24x7 cover provided by hosting location p Managed by the consortium 28
- 29. Layer 2 Exchange p Configuration n Recommendation: Public address space for IXP LAN p IPv4 (/24) and IPv6 (/64) n ISPs require AS, basic IXP does not p Network Security Considerations n LAN switch needs to be securely configured n IXP Management & Services router(s) require well protected access n IXP services must be behind router(s) with strong filters 29
- 30. Layer 2 Exchange p Industry Standards documented by Euro-IX, the European IXP Association n Contributed to by the Euro-IX members n https://www.euro-ix.net/starting-an-ixp p IXP BCP n General overview of the infrastructure, operations, policies and management of the IXP n https://www.euro-ix.net/euro-ix-bcp p IXP Website BCP n https://www.euro-ix.net/website-bcp 30
- 31. “Layer 3 Exchange” Why this is not an IXP 31
- 32. “Layer 3 IXP” p Layer 3 IXP today is marketing concept used by Transit ISPs n Some incumbent telcos call their domestic or international transit businesses “Exchanges” p Real Internet Exchange Points are only Layer 2 n L2 is the accepted International standard 32
- 33. “Layer 3 IXP” – what breaks p One extra AS hop between peers n Makes path via IXP suboptimal/less preferred n Path between peers usually remains with upstream transit provider p Unless both peers actively implement BGP policies to prefer the L3 IXP p Members cannot peer with whom they please n Mandatory multilateral peering n Third party (L3 IXP operator) required to configure peering sessions and peering policy 33
- 34. “Layer 3 IXP” – what breaks p More complicated troubleshooting n Troubleshooting peering problems has to involve IXP operator too p No policy control n BGP attributes shared between members gets dropped by IXP router n (Examples are BGP communities, MEDs) 34
- 35. “Layer 3 IXP” – what breaks p CDNs won’t join n They have requirements to peer directly with IXP members p Redundancy problems n L3 IXPs with dual sites appear as two separate transit providers between peers n Traffic engineering? p L3 IXP Operator requires strong BGP skills 35
- 37. Exchange Point Design p The IXP Core is an Ethernet switch n It must be a managed switch n It must have reasonable security features n https://www.euro-ix.net/ixp-wishlist has more details p Has superseded all other types of network devices for an IXP n From the cheapest and smallest managed 12 or 24 port 10/100 switch n To the largest switches now handling high densities of 10GE, 40GE and 100GE interfaces 37
- 38. Exchange Point Design p Each ISP participating in the IXP brings a router to the IXP location n Note that with increased availability of fibre access, ISPs connect directly to the IXP without provisioning a dedicated router at the IXP location p Router needs: n One Ethernet port to connect to IXP switch n One WAN port to connect to the WAN media leading back to the ISP backbone n To be able to run BGP 38
- 39. Exchange Point Design p IXP switch located in one equipment rack dedicated to IXP n Also includes other IXP operational equipment p Routers from participant ISPs located in neighbouring/adjacent rack(s) p Copper (UTP) connections made for 10Mbps, 100Mbps or 1Gbps connections p Fibre used for 1Gbps, 10Gbps, 40Gbps or 100Gbps connections 39
- 40. Peering p Each participant needs to run BGP n They need their own AS number n Public ASN, NOT private ASN p Each participant configures external BGP directly with the other participants in the IXP n Peering with all participants or n Peering with a subset of participants 40
- 41. Peering (more) p Mandatory Multi-Lateral Peering (MMLP) n Each participant is forced to peer with every other participant as part of their IXP membership n Has no history of success — the practice is strongly discouraged p Multi-Lateral Peering (MLP) n Each participant peers with the other participants (usually via a Route Server) p Bi-Lateral Peering n Participants set up peering with each other according to their own requirements and business relationships n This is the most common situation at IXPs today 41
- 42. Types of Operator Peering Policies p Open Peering n Where an ISP publicly states that they will peer with all parties who approach them for peering n Commonly found at IXPs where ISP participates via a “Route Server” p Selective Peering n Where an ISP’s peering policy depends on the nature of the operator who requests peering with them n At IXPs, operator will not peer with the “Route Server” but will only peer bilaterally p Closed Peering n Where an ISP decides who its peering partners are, and is generally not approachable to creating peering opportunities 42
- 43. Operators Peering Activities p The Peering Database documents ISPs peering policies and contact information n http://peeringdb.com p All operators of ASNs should register in the peeringdb n All operators who are considering peering or are peering must be in the peeringdb to enhance their peering opportunities p Participation in peering fora is encouraged too n Global Peering Forum (GPF) n Regional Peering Fora (European, Middle Eastern, Asian, Caribbean, Latin American) n Many countries now have their own Peering Fora 43
- 44. Routing p ISP border routers at the IXP must NOT be configured with a default route or carry the full Internet routing table n Carrying default or full table means that this router and the ISP network is open to abuse by non-peering IXP members n Correct configuration is only to carry routes offered to IXP peers on the IXP peering router p Note: Some ISPs offer transit across IX fabrics n They do so at their own risk – see above 44
- 45. Routing (more) p ISP border routers at the IXP should not be configured to carry the IXP LAN network within the IGP or iBGP n Use next-hop-self BGP concept p Don’t generate ISP prefix aggregates on IXP peering router n If connection from backbone to IXP router goes down, normal BGP failover will then be successful 45
- 46. Address Space p Some IXPs use private addresses for the IX LAN n Public address space means IXP network could be leaked to Internet which may be undesirable n Because most ISPs filter RFC1918 address space, this avoids the problem p Most IXPs use public addresses for the IX LAN n Address space available from the RIRs via specific RIR policies n IXP terms of participation often forbid the IX LAN to be carried in the ISP member backbone p IXPs provide both IPv6 and IPv4 support on IX LANs 46
- 47. Autonomous System Numbers p IXPs by themselves do not require ASNs n Ethernet switch is L2 device, and does not run BGP p Some IXPs have a Route Collector n This usually runs in a private ASN p Some IXPs have a Route Server n This usually runs in a public ASN p Some IXPs have “common good services” n These usually require Internet transit n Meaning the IXP requires a transit router p IXP arranges transit for services with a couple of providers n And this transit router requires a Public ASN and Public Address space 47
- 48. Hardware p Ethernet switch needs to be managed n Including CLI access rather than only SNMP n Unmanaged switches mean an unmanageable IXP p Insist that IXP participants connect a router (L3) port to the IXP switch n Avoid spanning tree and L2 security issues n Run port security or MAC filtering to protect the IX p Insist that IXP participants bring their own router n Moves buffering problem off the IXP switch n (Fibre access to IX reduces this requirement) n Security of ISP connection is responsibility of the ISP, not the IXP 48
- 49. Charging p IXPs needs to be run at minimal cost to its member participants p Common examples: n Datacentre hosts IX for free n IX operates cost recovery n Different pricing for different ports p IXes do NOT charge for traffic crossing the switch fabric n They are a peering enabler, encouraging as much traffic as possible between members 49
- 50. Charging: Datacentre hosts IX for free p Datacentre covers all costs relating to the IX n They provide the switch and supporting infrastructure n They provide the operator cover n They benefit from the business the IX members and their customers bring to the DC n They benefit from the “prestige” of hosting the IX and its ancillary services p The IX does not charge members for anything at all n Example: Seattle IX 50
- 51. Charging: IX Members pay flat fee p Each member pays a flat annual fee towards their IX membership p How it works: n Cost of switch and ports n Cost of operator support n Datacentre cost: power, air-conditioning, etc n Cost of IX membership association n Contingency needed for new equipment and upgrades p Total annual cost shared equally amongst members n The more members, potentially the lower the costs for each 51
- 52. Charging: Differential pricing by port p IXP Member pays according to the port speed they require n One linecard may handle 4 100GE ports n Or one linecard may handle 24 10GE ports n Or one linecard may handle 96 1GE ports n 96 port 1GE card is tenth price of 24 port 10GE card n Relative port cost is passed on to participants n Plus share in the cost of the switch n Plus all the costs mentioned in the flat-fee model p IX members pay according to the cost of provisioning their port speed n Example: Netnod IXes in Sweden 52
- 53. Services Offered p Services offered should not compete with member ISPs (basic IXP) n e.g. web hosting at an IXP is a bad idea unless all members agree to it p IXP operations should make performance and throughput statistics available to members n Use tools such as LibreNMS to produce IX throughput graphs for member (or public) information 53
- 54. Services to Offer p Root server n Anycast instances of F, I and L root nameservers are present at many IXes p ccTLD DNS n The country IXP could host the country’s top level DNS n e.g. “SE.” TLD is hosted at Netnod IXes in Sweden n Offer back up of other country ccTLD DNS p gTLD DNS n .com & .net are provided by Verisign at many IXes 54
- 55. Services to Offer p Route Server n Helps scale IXes by providing easier BGP configuration & operation for participants with Open Peering policies n Technical detail covered later on p Looking Glass n One way of making the Route Server routes available for global view (e.g. www.traceroute.org) n Public or members only access 55
- 56. Services to Offer p Content Redistribution/Caching n Various providers offering content distribution services n Broadcast media p Network Time Protocol n Locate a stratum 1 time source (GPS receiver, atomic clock, etc) at IXP p Routing Registry n Used to register the routing policy of the IXP membership (more later) 56
- 57. Introduction to Route Collectors What routes are available at the IXP? 57
- 58. What is a Route Collector? p Usually a router or Unix system running BGP software p Gathers routing information from service provider routers at an IXP n Peers with each ISP using BGP p Does not forward packets p Does not announce any prefixes to ISPs 58
- 59. Purpose of a Route Collector p To provide a public view of the Routing Information available at the IXP n Useful for existing members to check functionality of BGP filters n Useful for prospective members to check value of joining the IXP n Useful for the Internet Operations community for troubleshooting purposes p E.g. www.traceroute.org 59
- 60. Route Collector at an IXP 60 Route Collector R1 R3 R5SWITCH R2 R4
- 61. Route Collector Requirements p Router or Unix system running BGP n Minimal memory requirements – only holds IXP routes n Minimal packet forwarding requirements – doesn’t forward any packets p Peers eBGP with every IXP member n Accepts everything; Gives nothing n Uses a private ASN n Connects to IXP Transit LAN p “Back end” connection n Second Ethernet globally routed n Connection to IXP Website for public access 61
- 62. Route Collector Implementation p Most IXPs now implement some form of Route Collector n Usually as a Route Server (see next section) p Benefits already mentioned p Great public relations tool p Unsophisticated requirements n Just runs BGP 62
- 63. Introduction to Route Servers How to scale IXPs 63
- 64. What is a Route Server? p Has all the features of a Route Collector p But also: n Announces routes to participating IXP members according to their routing policy definitions p Implemented using the same specification as for a Route Collector 64
- 65. Features of a Route Server p Helps scale route distribution for IXPs n Forwarding of packets is unaffected n Makes use of BGP functionality known as “third party next-hop” p Simplifies Routing Processes on ISP Routers p Optional participation n Provided as service, is NOT mandatory p If traditional router used, will result in insertion of RS Autonomous System Number in the AS Path n To be avoided p Optionally uses Policy registered in the Internet Routing Registry 65
- 66. Diagram of N-squared Peering Mesh p For large IXPs (dozens for participants) maintaining a larger peering mesh becomes cumbersome and often too hard 66
- 67. Peering Mesh with Route Servers p ISP routers peer with the Route Servers n Only need to have two eBGP sessions rather than N 67 RS RS
- 68. Route Server based Exchange Point Routing Flow 68 TRAFFIC FLOW ROUTING INFORMATION FLOW RS
- 69. Using a Route Server: Advantages p Advantageous for large IXPs n Helps scale eBGP mesh n Helps scale prefix distribution p Separation of Routing and Forwarding p Simplifies BGP Configuration Management on ISP routers n Don’t need to maintain a large number of eBGP peers n eBGP peering only with the Route Server 69
- 70. Using a Route Server: Disadvantages p ISPs can lose direct policy control n If RS is the only peer, ISPs have no control over who their prefixes are distributed to p (Okay if ISP has Open Peering Policy though) p Completely dependent on 3rd party n Configuration, troubleshooting, reliability, etc… p Possible insertion of RS ASN into routing path n (If using a router rather than a dedicated route-server BGP implementation) n Traffic engineering/multihoming needs more care 70
- 71. Typical usage of a Route Server p Route Servers may be provided as an OPTIONAL service n Most IXPs now offer a Router Server as a service to members p ISPs peer: n Directly with significant peers -and- n With Route Server for the rest p ISPs with an Open Peering Policy usually prefer to peer with a Route Server 71
- 72. Route Server implementations p Linux/FreeBSD server: n BIRD – the standard & works best p http://bird.network.cz n Quagga p http://www.nongnu.org/quagga/ p Router: n Any router (but has RS AS in the AS-path) n Cisco IOS 15.2 and IOS XE 3.7 onwards has route-server-client configuration: 72 neighbor 172.16.1.1 route-server-client
- 73. Things to think about... p Would using a route server benefit you? n Helpful when BGP knowledge is limited (but is NOT an excuse not to learn BGP) n Avoids having to maintain a large number of eBGP peers 73
- 74. What can go wrong… The different ways IXP operators harm their IXP… 74
- 75. What can go wrong? Concept p Some Service Providers attempt to cash in on the reputation of IXPs p Market their Internet transit services as “Internet Exchange Point” n “We are exchanging packets with other ISPs, so we are an Internet Exchange Point!” n So-called Layer-3 Exchanges — they really are Internet Transit Providers n Router(s) used rather than a Switch n Most famous example: SingTelIX 75
- 76. What can go wrong? Financial p Some IXPs price the IX out of the means of most providers n IXP is intended to encourage local peering n Acceptable charging model is minimally cost-recovery only p Some IXPs charge for port traffic n IXPs are not a transit service, charging for traffic puts the IX in competition with members n (There is nothing wrong with charging different flat fees for 100Mbps, 1Gbps, 10Gbps etc ports as they all have different hardware costs on the switch.) 76
- 77. What can go wrong? Competition p Too many exchange points in one locale n Competing exchanges defeats the purpose p Becomes expensive for ISPs to connect to all of them n So they don’t, or won’t, and local traffic suffers, defeating the viability of the IXPs p An IXP: n is NOT a competition n is NOT a profit making business 77
- 78. What can go wrong? Rules and Restrictions p IXP tries to compete with their membership n Offering services that ISPs would/do offer their customers n In reality, IXPs are operated by the members for the members p IXP is run as a closed privileged club e.g.: n Restrictive membership criteria n In reality, a participant needs to have an ASN and their own independent address space p IXP located in a data centre with restricted physical/transmission access n IXP must be a neutral interconnect in a neutral location 78
- 79. What can go wrong? Rules and Restrictions p IXP charges for traffic n So do transit providers – charging for traffic is a sure way of ending the viability of the IXP p IXPs providing access to end users rather than just Network Operators & Service Providers n A participant at an IXP needs to have their own address space, their own ASN, and their own transit arrangements p IXPs interfering with member business decisions n The most common error: Mandatory Multi-Lateral Peering 79
- 80. What can go wrong? Technical Design Errors p Interconnected IXPs n IXP in one location believes it should connect directly to the IXP in another location n Who pays for the interconnect? n How is traffic metered? n Competes with the ISPs who already provide transit between the two locations (who then refuse to join IX, harming the viability of the IX) n Metro interconnections work ok 80
- 81. What can go wrong? Technical Design Errors p ISPs bridge the IXP LAN back to their offices n “We are poor, we can’t afford a router” n Financial benefits of connecting to an IXP far outweigh the cost of a router n In reality it allows the ISP to connect any devices to the IXP LAN — with disastrous consequences for the security, integrity and reliability of the IXP 81
- 82. What can go wrong? Routing Design Errors p Route Server mandated n Mandatory peering has no history of success n ISPs have no incentive to learn BGP n Therefore have no incentive to understand peering relationships, peering policies, &c n Entirely dependent on operator of RS for troubleshooting, configuration, reliability p RS can’t be run by committee! p Route Server is designed to assist with scaling peering at IXPs 82
- 83. What can go wrong? Routing Design Errors (cont) p iBGP Route Reflector used to distribute prefixes between IXP participants p Claimed advantages: n Participants don’t need to know about or run BGP n Allows an IXP to be started very quickly n IXP operator has full control over ISP activities n ISP participants routers sit inside IXP’s ASN p All are disadvantages! n Participants never learn BGP n Participants have no policy control, IXP policies could impact the participants networks n IXP is an ethernet switch, not an Internet operator n IXP operator is single point of failure n Migration to true IXP with RS is very difficult 83
- 84. What can go wrong: Summary p Not a transit business, just an L2 switch p If charging, fair cost recovery only p Not a competitive service p No oppressive rules & restrictions p No Mandatory Peering p No bureaucratic management p No interconnection with other IXPs p No bridging of IX LAN back to members p No Route Reflector, use a Route Server to scale 84
- 86. Exchange Point Policies & Politics p AUPs n Acceptable Use Policy n Minimal rules for connection p Fees? n Some IXPs charge no fee n Other IXPs charge cost recovery n A few IXPs are commercial p Nobody is obliged to peer n Agreements left to ISPs, not mandated by IXP 86
- 87. Exchange Point etiquette p Don’t point default route at another IXP participant p Be aware of third-party next-hop p Only announce your aggregate routes n Read RIPE-399 and RIPE-532 first www.ripe.net/ripe/docs/ripe-399 www.ripe.net/ripe/docs/ripe-532 p Filter! Filter! Filter! 87
- 88. Exchange Point Examples p LINX in London, UK p TorIX in Toronto, Canada p AMS-IX in Amsterdam, Netherlands p SIX in Seattle, Washington, US p PA-IX in Palo Alto, California, US p JPNAP in Tokyo, Japan p DE-CIX in Frankfurt, Germany p HK-IX in Hong Kong … p All use Ethernet Switches 88
- 89. Features of IXPs (1) p Redundancy & Reliability n Multiple switches, UPS/Generator p Support n NOC to provide 24x7 support for problems at the exchange p DNS, Route Collector/Server, Content Caches & NTP servers n ccTLD & root servers n Content caches n Content redistribution systems n Route Collector – Routing Table view 89
- 90. Features of IXPs (2) p Location n Neutral, secure & accessible co-location facilities p Address space n Public address for Peering LAN n Public address for IXP Services LAN p AS Number n Private ASN needed for Route Collector/Server n Public ASN needed for IXP Services p Route servers (for larger IXPs) p Statistics n Traffic data – for membership 90
- 91. IXP Creation p No economy or circumstance is unique or different n The first excuse for not creating an IXP is “we don’t need one” n The second excuse for not creating an IXP is “oh, it is different here” p Every locality has its differences n But every locality wants to p Keep local traffic local p Improve network performance and QoS p Improve local Internet economy n The available technology is the same for every network operator everywhere n There is no excuse for not improving the local Internet 91
- 92. Eco System Development p Create IXP association n Formed by members who have a port on the IXP p IXP members meet regularly n IXP Board meetings n IXP Operational strategy and direction p IXP Technical community could also meet too n Network operators meeting, involving network and systems operations technicians & engineers n Aligned with IXP Association/member meetings n Could lead to creation of a Network Operators Group p IXP could facilitate the creation of a NOG n The same technicians & engineers are involved in both! 92
- 93. Industry Associations p Euro-IX n European Internet Exchange association n Members from Europe, associate members from around the world n Website has all the information needed to start an IXP n https://www.euro-ix.net/starting-an-ixp n IXP Best Practice documentation: n https://www.euro-ix.net/euro-ix-bcp p APIX n Asia Pacific Internet Exchange association n Members from Asia Pacific region n Meets twice a year, during APRICOT and APNIC conferences n http://apix.asia 93
- 94. More info about interconnects p Telegeography n http://www.telegeography.com/telecom-resources/ internet-exchange-map/ n A collection of ISP interconnect points n Beware!! Not all Telegeography listings are IXPs! p Internet Society n http://www.internetsociety.org/what-we-do/issues/ internet-exchange-points-ixps n http://www.ixptoolkit.org/ 94
- 95. Summary p IXP is a Layer 2 infrastructure p At least three players required (two is okay too) n Meeting in an open and neutral location p Minimal rules p Minimal bureaucracy p Cost recovery p Encourage participation by all autonomous networks p Develop the local Internet eco-system 95