autonomous system routing protocols | BGP | autonomous system number | internet exchange point

in our first session we talked about the Internet as being a network of networks and those networks are generally organized into what are called autonomous systems the autonomous systems are designated by provider such as 80 in this case we see two autonomous systems one for the US and one for the Latin America and they are really standalone networks that enforce the IP protocol at the edges but internally they're managed by that single administrator and designed by that particular provider the number the autonomous system number and we see three of them here are assigned by the regional internet registry under the rules in the pools that are set by I Anna there are both public and private autonomous systems just is there a public and private IP addresses and the private autonomous systems generally handle quote enclosed traffic that is wall garden networks or private virtual private networks is in them the reviews for autonomous private autonomous systems private autonomous systems can use the same physical network in the same routers as the public autonomous systems which serve the Internet there are three methods of interconnecting autonomous systems and from now on my comments will apply to the public Internet and the public autonomous systems the first we've already talked about settlement free peering and settlement free peering is conducted between Tier one networks but it's also performed between a company like AT&T and their Tier one and tier two networks in Latin America eighteen t is a Tier two provider and so they interconnect the Tier one and the US with that Latin American network that's really a bridge more than it is appearing is it as it is in Tier one so we'd see a variety of traffic going across that a Tier one tier two network that AT&T would provide its customers a second kind of peering is private peering now private doesn't refer to Private Networks but rather the fact that the arrangement between the two networks AAS one two three four and eighty in this case is done on a negotiated basis so the arrangements in the settlement fees are set between the two it is for public Internet traffic however and this case the settlement agreement could mean that a tea only provides connection to other tier 1 providers and other ISPs and content delivery networks that are directly connected to the tier 1 it might not include porting traffic and receiving traffic from the tier 2 network in Latin America that's all a function of the arrangement between the companies and private peering a third and very popular kind of interconnection is public peering in public peering occurs at inter exchange points IXPs those used to be called network access point naps but the name has changed on XP and I XP is generally run by a non-profit group that is facilitating exchange of traffic among networks and so it essentially looks like a large layer to a network that provides layer 2 connectivity between all of these companies in the hosting in the in exchange point we would see hosting companies like Amazon and GoDaddy and Rackspace coming there with facilities that would connect it as one physical location we would also see content delivery companies like Akamai and limelight and would see ISPs those are all connected as well as Tier one networks at particular physical locations around the world these companies like the hosting company in the US have a choice between buying facilities to get to the IXP or contracting on a private peering basis with a peer one provider like AT&T in the next slide we're going to see how these interconnections actually function from a protocol viewpoint in my example I'm going to be a user sitting in an autonomous system 1 2 3 4 could be an ISP or it could be at an enterprise that's contracted with AT&T for Internet service and I want to access a website in Japan and that website is for orchids for exotic flowers and orchids in particular so we're going to go through first of all the steps that the website administrator would have to go through to put their site on the network and get it advertised and then we'll go through what a connection looks like we first have to build connectivity so orchid that is the company that once they're content on a website has to select a hosting ISP and then get an address from that ISP the ISP gets the address blocks from the regional internet registry in this case it's the asia-pacific registry which is called ethnic as a side note ipv4 addresses were exhausted in Asia Pacific about a year ago so this would have to be from a pool that the ISP has and is not exhausted otherwise you're going to have to go to ipv6 which we'll talk about a little later the next thing that needs to happen once that address is allocated and here's the IP address it has to be registered in an authoritative DNS that's done so that all the other DNS is on the Internet can find the IP address through the process that was described in the previous session now the address has to be advertised the DNS lookup that's going to be performed when I try to access that website is only going to return the destination IP address it's not going to tell me how to reach it the protocol is used to advertise addresses are called border gateway protocols and we're going to examine two of them here one is an external border gateway protocol so that is a protocol for exchanging address availability across a s's it is not a routing protocol it's strictly used to expose addresses in identify where connectivity or reach ability I should say is possible so in this example a s 2914 would send an e BGP message to a s 7 o 18 at the peering router that would have its own s in the message as well as in this case I'm going to say a subnet 203 183 and 38 a BGP is generally not used for individual addresses but range of addresses and in fact it would probably be a much wider range than shown here but that's what I'm going to use in our example the bgp update only brings the information to the edge router the as7 o-18 has to update all the other edge routers within its network so that all of the edge routers are in saint's realization and they all know the edge router that is connected to this particular IP address and this would be our egress router so this is our egress router the ingress router will be a function of where the incoming traffic is received so now we've updated the survey the egress edge router and updated all the internal edge routers in the AT&T network the edge routers in turn update all of their ebgp peers that is all of the CER outers and all the inter connecting routers that accept the BGP updates and now the address is advertised into our one two three four autonomous system and the the ebgp protocol so we've had three sets of protocol updates in order to get this information over there the eye BGP is done using the TCP protocol as is the e BGP and TCP is used because it can be done across networks and it's not a hop-by-hop arrangement it's an in the end arrangement and it's reliable so now we'll look at how the IP packet is actually for tea first a DNS lookup has to be performed and because Orkut has been loaded in an authoritative name service DNS that serves our host machine will locate that record that resource record and return it and resolve it and we describe that in the last session the CEO outter will form receive an IP PE u from the host machine is the default gateway and because BGP has notified it that this address is in fact reachable via the AT&T network it will afford that IP PU to the 701 8 edge so that would be 42 here now the PE router because of ibgp knows that this peering or bridge router is the egress so it knows the egress router is right here so it's going to afford that IP PDU to this edge router fording occurs in a way that's specified by the autonomous system and there's no particular requirement for to use one technology over another the most common scalable method of 14 in these large tier 1 networks is a protocol called multi-protocol label switching with MPLS the there's a layer below the IP layer called the MPLS layer which forwards the IPP U and it does not use a hop-by-hop OSPF forwarding but rather a hop-by-hop MPLS 14 based on tag switching so I won't get into that here and a great deal but from an IP view point of the PE router sees the egress router is the next hop MPLS handles the mechanics of getting the packet from the ingress router which is here is the ingress router to the egress router at the egress router the IP PDU is forwarded to AAS 2914 and so now our packet has been delivered to the hosted destination there's a ongoing really update that needs to be done in an MPLS Network and the internal label distribution is performed at each of these P routers so we have provider edge router in P routers which are the core routers the MPLS updates are done by the MPLS protocol and by a label distribution protocol so the labels tables that is the tables that specify how you handle MPLS routing is distributed using an internal protocol OSPF would be one choice is is would be another that's more than we have time for this in this session but it's a it's worth noting that there are other protocols operating besides just the bgp protocols and MPLS inside these networks while this really design looks quite complicated it's really quite effective and it is effective because it scales extremely well these networks scale to hundreds or even thousands of routers and the reason it scales is that each network element only has a limited responsibility these two edge routers are primarily concerned with reach ability in forwarding across the boundaries they only need to know about other addresses in the network as they're being informed by ibgp the core routers which are large-scale routers do not need to update their routing tables at all as regards to the external networks because it is really gateway edge routing protocol that performs the final resolution and advertises a reach ability the core routers are really routed on MPLS and of course the internal OSPF routing in order to update the labels and other parameters and information within the core so this is a great example again of segregation of responsibilities in order to be able to scale in the next slide I'll show you one [Music]

Channel: ISO Training Institute

Views: 7,524

Rating: 4.4418607 out of 5

Keywords: autonomous system in networking, autonomous system number, autonomous system bgp, autonomous system eigrp, autonomous system routing, routing protocol, connecting autonomous system, internet exchange point, internet exchange, IXP, autonomous system number in networking, ip forwarding, dns lookup, autonomous system, ccnp, bgp, peering, internet autonomous system, autonomous systems, eigrp, routing basics, open shortest path first, routing protocols, border gateway protocol, routing

Id: C8f9KCGOBAs

Length: 15min 10sec (910 seconds)

Published: Fri Mar 10 2017