Private peering makes sense when there are very few equivalent players. Connecting to one other ISP costs X. Connecting to two other ISPs cost two times X. Connecting to three other ISPs cost three times X and so on where X is usually half the cost of infrastructure the circuit plus a port cost the more private peers the greater the cost and this doesn't scale the costs go up as we add each private peering link the internet exchange point is a more scalable solution to this problem connecting to an exchange point means the ISP pays for one router port on the local border router one circuit to where the exchange point location is and potentially one router to locate at the exchange point some exchange points will charge and your maintenance fees this goes towards the cost of the infrastructure the air conditioning the remote hands and so forth the maintenance fee actually has potential to significantly influence the cost balance for a service provider generally connecting to an exchange point in pealing there becomes cost effective when there at least three other peers but the real amount varies from region to region from exchange point to exchange part so who would peer at an exchange points access providers would go there because then they don't have to pay the regional provider transit fees for local traffic this keeps latency low and it keeps the costs for local traffic low as well and this basically unlimited bandwidth through the exchange part compared with costly and limited bandwidth through the transit provider regional providers would also peer at an exchange part because they don't have to pay the global provider transit fees for local and regional traffic and likewise it keeps latency for local and regional traffic low and it keeps the costs for local and regional traffic flow and again thus unlimited ban word through the exchange point at least when compared with costly and limited bandwidth through the global provider we also find that content providers and content distribution services peer at exchange parts they don't have to pay the regional provider for transit fees for local traffic again it helps latency and cost for local traffic and again they see the unlimited bandwidth through the exchange point infrastructure for the same reason we will find the root name server operators country top-level domain operators and the general top-level domain operators this adds resiliency to the global DNS system as well as keeping latency and response time for local resolver traffic very little this effectively speeds up the DNS and gives the end users of the impression that the internet is fast continue the discussion about the exchange points role we also find that global providers will be located close to the exchange points they can be attracted by the potential transit business available they won't peer at the exchange part themselves but will be located close by and this makes it advantageous for the access and regional providers they can pair with other similar providers at the exchange point and in the same facility they can pay for transit to the regional or global provider they don't take the transit across the exchange point fabric most exchange points don't allow this and anyway it's not good practice they generally provide a separate private connection back to our P&I to connect to the transit provider now let's look at the connectivity decisions for transit provider you need that to reach the rest of the internet so every ISP does need a transit provider one provider is no redundancy two providers is ideal for traffic engineering as well as redundancy three providers gives even better redundancy but we have to work a little bit harder with our traffic engineering more than three provides diminishing returns rapidly escalating costs and complexity and is not recommended by most major operators today peering means low or zero cost access to another network and peering can be private or public as we've learned earlier the goals of an operator are these to minimize the number of transit providers but maintain redundancy to is ideal for or more is getting really hard also we want to aggregate capacity to our transit providers the more aggregated capacity there is the better the value we see because of the law cost per megabit per second as the bandwidth capacity grows for example for stm-1 or OC 3 links which 155 megabits each to four different ISPs will almost always cost more than two stm-4 or OC 12 which is 622 megabits links to two different ISPs yet the bandwidth of the latter 1.2 gigabits per second is greater than that of the former 620 megabits per second is much easier to configure as far as bgp traffic engineering goes and is much easier to operate in the medium to long term so what do we do we obviously need peering and we need transit how do we choose though and it comes down to the cost of going to an internet exchange point at an exchange point we will get free peering and we could pay for transit from an isp which is cool located in the same facility or maybe close by it could be in a neighboring building or in the same city or we could just decide not to go to an exchange point and pay for the cost of transit directly to an upstream provider there may be no exchange point in the vicinity and so transit might be the only option there's no right or wrong answer to this someone has to sit down and do the arithmetic work out what the costs are should we be doing private peering well as we've learned there's a scaling issue the more circuits we have two more neighboring providers the greater the cost will become one about public peering well this will make sense when more potential peers are located in the same location and more is usually greater than two and then there's a question about which public peering point we should go to well the local internet exchange point is great for local traffic and local peers and there's almost no argument for not turning up at one. Regional internet exchange point is also interesting. It's great for meeting peers outside the locality and might be cheaper than buying transit to reach the same consumer base.

Related Documents

Exercises

© Produced by Philip Smith and the Network Startup Resource Center, through the University of Oregon. 

Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)

This is a human-readable summary of (and not a substitute for) the license. Disclaimer.
You are free to:

Share — copy and redistribute the material in any medium or format
Adapt — remix, transform, and build upon the material
The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms:

Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
NonCommercial — You may not use the material for commercial purposes.
No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.