400 101 CCIE Routing and Switching 94 IP Routing

[Music] the time has come to actually start talking about IP routing so when we talk about IP routing for the purposes of the CCI routing and switching exam basically what we care about or some of the protocol independent features and these are things like the admin distance the network statement which kind of is protocol dependent but still I like to talk talk about it from the protocol independent point of view we are going to talk about the passive interface concept and then of course we are going to talk about different routing protocols and here while protocols again with an asterisk there we're going to talk about the connected route you're going to talk about static routes we are going to talk about rip EA GRP or SPF and of course bgp even though I like to talk about BGP separately from the others because it is kind of a special world in its own and I like to actually think of BGP as not being actually a routing protocol I like to think of it more of a policy implementation protocol so let's start this over now the first thing that I'm going to talk about is the administrative distance now we all know what the admin distance is it's a tie breaker I really need to work on my handwriting color it's a tiebreaker between routing sources so if we have on one router oh let's do that properly routers are always round so if we have a router and this router here has multiple sources of routing information so the information is coming from multiple protocols and let's say that from all of these sources we are learning the same route let's say 192 168 0 0 slash 16 this router somehow needs to figure out which one of these sources is more preferred than the others so this is where the admin distance is going to be the tiebreaker before this router can actually install the route from one of these sources in to the routing table now the important thing to note here is that the admin distance will be used only in situation when you have the route coming from different sources of routing it will never be used as a tiebreaker inside the routing protocol so for example inside OS we have inside a single OSPF process if you have the same route learn from multiple therefore multiple sources and if you try to modify the admin distance of one or the other you're actually modifying the admin distance at this point here after the protocol has actually sent the route to the routing table you cannot use the admin distance to influence the decision made by the protocol itself the reason for this is that the admin distance is a vendor proprietary proprietary concept cisco calls it the admin distance HP Dell and others are calling it the admin distance juniper is calling it the routing preference but it's really really the same concept it's the preference given to a particular routing source but not the influencer inside the routing protocol itself because the routing protocols are usually defined in standard documents like for example RFC's or Cisco internal documents that specify how ear GRP behaves or nowadays actually the RFC that specifies how ear GRP behaves because it is well since relatively recently a standards-based protocol surprise now different routing sources in Cisco IOS have different routing preference given so for example the most preferred routes are of course the connected routes now the connected routes have the admin distance of 0 and this is not something that can be changed so this is something that is fixed static routes are by default the second most preferred routing source and they have the admin distance of one and this can actually be changed now the next preferred routing source our ERP summary routes and these have admin distance of 5 this also can be changed so let me talk a little bit about this GID RP summary Rob because this is one thing that I have seen my students from time to time have little confusion about so to do this I'm just going to use two routers in my mix I'm going to have our two directly connected to our five now our five here will have a loopback interface five five five five / 32 and the link here this is serial 0 to 0 and this is serial 0 to 0 and the link here between them is going to be 192 168 25 0 let me just very very quickly configure ERP 25 between these two devices so I'm going to go to my R 2 and R 5 now I do have some other interfaces configured right now so I'm quickly going to remove those interfaces so default interface serial 0 0 0 and on our to know interface loopback sir so interface loopback 0 IP address 5 4 5 5 and if I do show IP route connected I should be seeing this was already pre-configured so I should be able to ping 25/2 so this is already in place so next thing I'm going to do is it just going to quickly bring my text editor here and I'm going to say router ERP 25 no Altos summary Network I want to run a idrp really on all of the interfaces this is just going to be a very quick exercise because the point here is not about the EIGRP configuration the point here is to show you the eigrp summary so if I do show IP route EIGRP now on our - I'm seeing this five five five five with the admin distance of 90 so this is the default admin distance for ERP so maybe I should actually just add it here so for EIG RP the admin distance ya GRP internal routes they have admin distance of 90 this also can be changed using the configuration so now what I'm going to do is I'm going to configure on this interface here OHRP summary to summarize five zero zero zero slash eight so I'm going to go to my router r5 and on the interface serial zero two zero I'm going to say IP eigrp okay sorry IP summary address CRP twenty five five zero zero zero like this and here at the end what I can do is I can specify the administrative distance for this summary but if I press ENTER the administrative distance for this summary will be five now after my routers have actually reset if I go to r2 and if I do show IP route EIGRP I will be seeing this summary but take a look at the admin distance the admin distance on this route is still 90 so where is this admin distance of five that we were talking about well for that we actually have to take a look at our five so on our five if I do show IP route AI GRP take a look at this now on our five on the router on which I have actually created this summary so the result of this command is this GID our peer out on our five take a look what is the exit interface for it it is null 0 so this route here also called the discard route has been installed as the result of me creating the summary route so here at the end I didn't type this 5 I just pressed enter after a GRP summary dressy summer dressy IGR p25 5/8 this is where I pressed enter but iOS installs the this five by default so the result of this summary address is this discard route that is installed on our 5 as the EIGRP route in our 5s routing table now what is the purpose of this wrap if you try to explain now when we create a summary address so let's expand our example let's say that I have r2 here and let's say that I have our five and let's say that instead of having a loopback interface I may have some interface here that has multiple routers multiple devices and let's say that this network here this router here was advertising 5 5 0 0 / 24 this one here was advertising 5 5 1 0 / 24 this one here was advertising 5 0 3 0 / 24 etc now when all these routes were actually received by r5 I would have any number of these / 24 out now what I want to do here from r5 I want to tell our - hey you know what if you have any traffic that goes to 5 0 / 8 send it to me so let's now say that our - sends the traffic for far five seventeen nineteen thirty one okay so something that we don't have here in this network now when this traffic actually arrives to r5 as the result of this route because our - doesn't know any better our v is going to say okay I'm having traffic for 5:17 1931 do I have a more specific route where to send this traffic no I don't so r5 is actually going to drop this traffic as the result of this now route now imagine that our to here was actually the router that is connected to the Internet now in that case it is perfectly conceivable that r5 actually has a static for example default route pointing to r2 now if we had that situation and if we didn't have this null route installed sorry this route that will cause the traffic to be dropped here let's take a look at what would be the situation then so I'm going to go to my terminal and on r5 I'm going to say that you know what artoo is your default gateway to the Internet so this is the situation with the discard rod so let's go to r2 and let's try to ping 5:17 what was the address 1931 so I'm trying to pin it and obviously I don't have the path to it I see the time getting destination unreachable so if I do trace route to this destination I will see that traffic takes me to our five and then our five actually drops this traffic that's exactly what I wanted to have now let's on our five get rid of the discard route so what I'm going to do here is I'm going to go into the sub interface mode and I'm going to install the same EW GRP summary except that now I'm going to give it the admin distance of 255 now I'm in 255 admin distance to five five so let me put it here and I'm going to put it all the way at the end admin 255 is an untrusted source and this is a fixed value and this simply means that the route that have the admins distance of 255 will not be installed in the routing table so now let's take a look what will happen when I get rid of the discard route so the discard route is now gone so if I do show IP route most likely isn't gone but I know how to get rid of it so do clear IP route star so I'm just going to get rid of it now so if I do show IP route here GRP now the discard route is no longer in the routing table so from our two I'm going to try to ping 517 1931 and you can see the time now getting slightly different result it may not be immediately obvious but this is very very different result because take a look at here in this case I was actually sorry about that I was actually getting ICMP unreachable here the traffic simply doesn't seem to get to its destination if I do a trace route this is why because what is happening now is r2 is sending the traffic following the summary then our five says oh this is traffic for 5:17 1931 I don't have more specific route for it so but oh I do have the router it look it's the default route pointing to r2 so traffic now goes to R 2 and then r2 says oh but this is traffic four five zero zero zero slash eight and sends it to r r5 and repeat until the TTL expires the default TTL for trace routes is 30 so we will not go more than thirty Hobbs but theoretically this could this could circle between r2 and r5 up to 255 times so this is what the discard route is trying to prevent to do whenever we are summarizing IP information like we've done here what we are doing is we are losing information we are losing the granularity of the information so we somehow need to provide for the loop prevention in this case so this is what the discard route does so let me put it back in because it is actually a very useful thing to have so I'm going to put it back in and I'm going to have it with the admin distance of five so if I do show I threw out here GRP I have this route let's go back to pinging this address now now when I ping it I am getting these ICMP unreachable now this ICMP unreachable are sent by r5 when it needs to drop the traffic as the result of sending this traffic to null zero interface now if you are receiving a lot of this traffic that needs to be discarded that can put additional strain and load on our five and incidentally this is something that they may ask you to address in the lab so the question that they could ask you in the lab along these lines would be make sure that when are five discards the traffic that follows the summary that you are discarding this traffic without sending the IP unreachable traffic or IP ICMP unreachable so the question here is where would you put IP ICMP unreachable well let's try putting them on the serial interface so no IP ICMP unreachable hold on what am I doing wrong no oh no IP unreachable sorry so when I when I originally started saying no IP unreachable that was actually thinking about the IRS command but what this is sending is what this is disabling any ICMP unreachable you know sometimes there is disconnect in my head between what the features actually are and what the iOS command is I feel not disconnected my head disconnecting you know iOS but I digress so let's try to ping now let's see what happens now see the ICMP unreachable are now not being sent but what if our five had multiple interfaces here what if I had many interfaces where would I enable IP ICMP unreachable if I would have to do it on every single interface there is actually a more efficient way to do this so I'm going to go and re-enable IP ICMP unreachable so now I'm getting them and on our five instead of configuring the serial interface I'm actually going to configure interface null zero now or interface now 0 there is surprisingly little you can do pretty much the only thing that you can do is know IP unreachable so now when I try to ping I'm not going to get me getting IP and reachable and I don't actually have to worry on which one of these interfaces I have enabled it because I have actually enabled it on the discard interface on the interface that will be dropping the traffic this is where I am preventing the IP and ICMP unreachable from being sent from so this is what I wanted to show you when it comes to a to admin distance 5 now this functionality here so I'm just going to write it is called discard route now this card route is automatically created by AI GRP whenever you create a summary it's automatically created by OSPF when you create the summary and is automatically created bgp when you create the aggregate address now if you want to get rid of it in AI GRP it would be admin distance of 255 on summary in OSPF it will be no discard route under the OSPF process and for bgp sorry you will always have it so you cannot disable it for the aggregate address that is created now keep in mind that there is a notable absence of rip-rip does not create a discard route when it creates summary so this problem with the default route and the summary that we had between r2 and r5 on RIP is something that is unfortunately the fact of life so it is not something that you can fix with rip well you can manually you could add the static route that points to null 0 for the summary but in the CCI lab I want to have to have this mental picture when you type IP route something there is a piano falling on your head you know just like in the cartoons so type IP route and playing you are gone right so don't type static routes in the city I love unless you are very explicitly allowed to use the static routes generally speaking they are not allowed in the CCI lab so let's go back to up to this thing here the next best Edmund distance that you care about is of course going to be OSPF internal routes which are going to have the admin distance of 110 and of course this can be modified then the next thing that you care about is going to be rip with 120 and this of course also can be changed then we have e IG RP external routes with 170 and this of course can be changed and finally we have bgp externals I'm sorry oh I forgot - bgp externals miss those so I'll have to go back and fix that but anyways so BGP internal routes with admin distance of 200 so let's fix that missing BGP they are actually right here so we have BGP external routes I knew there was something missing there just wasn't thinking straight it's 20 and of course it can be modified now there is one here that you don't really care about but I'm going to write it anyways it is is is is sitting between OSPF and rip with 1 1 5 and of course it can be modified but because you don't have is is on your lab or ICS as I like to call it because you don't have a science you don't really care about this now this table these values is something that you should really really really know when you go to the CGI lab mostly because this is really the CCNA level administers so let's bring up the pace a little bit let's talk a little bit more about admin distance about I would like to say the CCNP level of admin distance now let's say that we have a router here and that we are learning so we are going to be learning the same route so this is 192 168 0 0 slash 16 so I'm going to be learning this route from two sources let's say that this is ripped which has the admin distance of 120 and this here is OSPF and this one has admin distance 120 as well now mind you 120 isn't OSPF 110 well yes it is but we said that this value can actually be changed so I have modified the admin distance of OSPF in my case to be 120 to make it the same as so I'm receiving this same route with admin distance of 120 and admin distance of 120 and let's say that the metric here was 5 and that the metric here was 200 which one do you think will be installed in the routing table now if your answer is rip please think again because metrics cannot be compared in this case the only thing that matters is the admin distance but admin distance is the same does this mean that we are going to load balance between these route there can be no load balancing between the traffic for or sorry for the traffic coming from two different sources so what we need to do here is to choose which one of these 192 168 0 16 is the best this is where the original admin distance will be compared the original hard-coded admin distance for rip this is 120 and for OSPF this is 110 so based on this value the original administers or the default or the hard-coded call it whatever you want to call it I like to call it the original but sometimes that confuses people so here I'm going to give you what I actually mean by that based on this information the OSPF route will actually win even though on face value their admin distances were the same so this is easy enough so if the current the actual electrical the operational admin distance is the same the original the hard-coded admin distance will be used as a tiebreaker but sometimes in some cases this is not actually going to work now what is this case going to be so let's say that we have a router again we have a router here and we have a route again the same route 192 168 0 0 slash 16 the teaser learned from two sources and let's say that this here is OSPF process one and this here is OSPF process - now the admin distance here is 110 the admin distance here is 110 let's say that the metric here is 100 and the thematic here is 200 so which one do you think wins now if your answer here was OSPF one think again you may or may not be right again the metrics cannot be compared why because these are two completely separate routing processes now think about it this way and this is a joke that I always like to do in my classes so if metrics of OSPF 1 and OSPF 2 can be compared I am now sitting in the United States of America where generally accepted officially which even though technically speaking it's not the official language is English language so I am speaking some version of English to you now English means it comes from England and England has a queen so given the fact that we are speaking the same language here in the United States as they speak there in the England or so in England that means that there is a queen of United States and her name is Queen Elizabeth the second the queen of the United States well obviously there is not some there is not any truth there in there unless you of course ask somebody from England and I'm looking at you you know who you are who may disagree with this statement but there is this tiny little thing that was called War of Independence that actually took care of not having the queen of the United States so this is the same situation here Oh SPF 1 and OSPF to think of them as United States and England right even though out the language is technically speaking the same there are very very different laws and different regulations and the laws in England cannot be compared with those laws in the United States even though some of them are probably the same I guess don't steal is one of them I don't think it's legal to steal unless you are high income banker in that case I guess that's just daily business for you but in this case the metrics cannot be compared so what can be compared well what we can pierre is the original admin distance well the problem there is that the original admin distance is exactly the same so this also won't help us make a determination now this is where the consistency of iOS strikes again for OSPF if you are using OSPF as a routing protocol and if you have this situation that you're learning the same route from two OSPF processes what is going to be the situation here is that the route learn first wins this is also true for is is and which you don't really care about but I think it's fun to mention it for EIG our Pido route from now this is where I always forget so let's test this it is going to be the route from either the lower autonomous system or the higher autonomous system but you know what and this illustrates the point you know I know so many things about network they're things that I don't know about networking there are many many things I don't know and there are things that I simply cannot remember no matter how many times I try them so every time I'm in a situation that I have a doubt about something I am actually going to live it up good news is that I actually have the lab that I can easily do that on so this is the setup that I have right now so let me add another router so I'm going to add our forth to the mix and I'm going to from are to advertise let's say 24 0 0 0 slash 24 network and from r4 I'm going to advertise 24000 slash 24 now here I'm going to run EW GRP 25 which incidentally is already running and here I'm going to run EA GRP 45 so I'm simply going to see which one of these two routes that are exactly the same will win on our five very very simple tasks so let me go to my terminal I'm going to add back our forth to the mix I already have no I actually deleted that interface so let's configure it so IP address one or two 168 45 five no shutdown and I may need to modify the process of AI GRP just a little bit so I'm going to say network one is to 168 25 0 no network 0 0 0 because I don't really want this process to run towards our 4 so here I have it now let's create that other process or outer ear GRP 45 no auto summary and you know what I can just copy paste this and change this to 4 so this will take care of it now one hour to let me create a interface loopback 24 IP address 2400 to two five two five two five zero that will advertise this interface on r5 so I can see here show IP route EIGRP there it is and let's go to r4 so on our 4 I'm going to say router let's create the interface interface loopback 24 IP address 2400 for 2005 0 Rodrigo GRP 45 no Auto summary and here it's octet to actually run on all the interfaces so now when I see the eigrp neighbor come up if I do show IP route eigrp I should be sick actually there is a loopback 0 here so no interface to talk Sarah let me get rid of that one so here I can see that the route from our two is actually winning in the routing table so if I go to interface serial 0 to 0 and I shut it down here I see that one from our 4 is now in the routing table so remember this process here has the higher value than this process here so it has the it's it's the higher a s value so if I bring that interface down no shutdown when interface comes up there it is an e IG RP I can see that our five now prefers the route from our two so to answer my own question my own doubt here it will be the route from the lower EA GRP yes number now while I'm at actually proving how things work let's see this one in action here as well again a very very simple test I'm going to use the exact same network so I'm going to have R 2 R 5 and R for here I'm going to have 2400 0/24 same thing 24000 / 24 and let's run OSPF 25 here and let's run boy SPF 45 here so pretty much the same setup I have right now I just have to change the routing protocol from ERP to OSPF so no router erp 25 I'm going to say rudder oh SPF 25 network area zero on our for no Rutter AGP 45 the rudder oh SPF 45 network area zero on our 5 no router ERP 25 no router ERP 45 and rudder oh SPF 25 Network 1 and 2 168 2500 and I will have our SPF 45 with the network 45 so now I can see that both of these processes are have loaded so show IP ospf interfaces brief I just want to see that I do actually have neighbors so I can see that different processes are running on these two interfaces and I can see that I have adjacent neighbor so if I do show IP route OSPF I see oh oh sorry loopback interfaces apologies there is an extra thing that needs to be done there and I will talk about this so apologies I just forgot that I need to do this otherwise they will be advertised as slash 30 tools and in that case they are not the same route so here now I see that from r2 was the route that I'm preferring so now what I'm going to do is I'm going to go on r2 and I'm actually going to shut down the interface between r2 and r5 so now our five is actually going to be preferring the route from our four if I do now shut down on the interface and when OSPF converges there we go change state two up it's changed from loading top four let's give it ten more seconds I will see the time still preferring the route from our four why because it was the older route now if this route disappears so interface serial zero one zero shut down now our five will actually prefer the route to r2 as you can see our two is now the route in the routing table now when this interface comes back up and when our four advertises the route again this route here is going to be considered the all the route because it was learned first so you can see that this one is still in the routing table so OSPF prefers the older route to new we're out in the case that you have this tiebreaker this is very important because OSPF prefer stability to lower a s number that ERG RP does but this is actually very dangerous situation because at this point when these routes are received from r2 and r4 when they need to be installed in the routing table of our five there is no apology differentiation between the internal route and the external route in OSPF to illustrate this what I'm going to do now is I'm going to go to our for and I'm going to say in the OSPF configuration so router OSPF this was 45 instead of run this on all networks what I'm going to do is I'm going to get a little bit more specific so one two one sixty eight forty five zero like so and I'm going to remove this Network statement here so this is the configuration now on my R 4 and then in the router OSPF process I'm going to say redistribute connected subnets so now on our 4 R 5 this is the route that is preferred mind you this is the route from R 2 n is an OS PF oops it's an OSPF internal route I can see that by oh here if I go to our tube interface 0 0 to 0 and I shut this down on our 5 what I'm going to have now maybe I need to do interface you know z10 know maybe the interface was shut up so if you're deaf neighbors nope wasn't so let me see why am I not redistribute in I can see with this we'd connected subnet why my nostril distributing this should have been redistributed but it wasn't over after did I configure the wrong process yes I was oh Jesus ok let me fix this so I'm going to say redistribute connected subnets and autopilot in the fingers Oh Roger OSPF one very very dangerous if you're using on one process you'll be configuring wrong stuff anyways now I do have the external route on our five now we know that OSPF prefers internal over extrude but what's important here is admittance 110 so now let's bring that interface up on r2 so now r2 is going to be advertising internal rod take a look at which route is in a routing table it is the external route now this is exactly what I was saying before at this point of comparison when r5 needs to choose between this process and this process it doesn't matter whether the route was internal or external or the what matters is only the admin distance this is 110 this is 110 if this is not comparable we are going to go for the original administers which is going to be 110 here and the original admin distance here of 110 if this is the same as it is the same we go for the oldest route and in this case the oldest route is the route from here which happens to be the external OSPF route inside this OSPF 45 process but this doesn't matter anymore here I have in the routing table OSPF external route which is preferred over an internal route well internal route from another process and remember England United States the internal here means nothing internal here the only thing that matters is the admin distance so the question here was what happens if I clear IP route okay let's do clear IP route star and see what happens this one wins why did it win well because at that time what happened was when a clearly IP route from the routing table the routing table is going to go to the processes OSPF on one side and OSPF on the other side and it's going to ask them hey give me your route and this is a noise this is an iOS internal process and it could happen that it asks the process towards our to first because in this case it has the lower processed number so in this case this might be a situation and let let's test that example as well so here what I'm going to do I'm going to remove the OSPF process 25 and I'm going to create a SPF process 250 that is going to have the exact same configuration so you ask me the question that made me really really curious I don't know answer to this one so that was my educated guess at this point so so IP out OSPF now the external one is the older one because I removed the process but if I do clear IP ospf process star that's sorry no clear IP route star now the routing table now goes for the route - rerouting process is telling them hey give me your best route so if I do show IP route OSPF well it looks like the internal in this case wins so the answer is I really don't know at this point what what could be the cause of this but you know what you made me so curious the time actually going to go and try to find the answer now the good place to find the answer for this particular behavior would be the book is called I believe iOS internals it's rather old but it could answer this question if the answer is not there I would have to grab someone from Cisco hint-hint if you are online to use something called topic to answer this question for me the topic is Cisco's internal corporate brain where the information could be found but this particular behavior here to be honest I expected the external route to to pop in here but let me let me try one more thing let me try one more thing let me try one more thing just want to try this it was a very very good question now the internal still wins in this case hmm very curious as I said don't know the answer to this particular one but I would have to uh I would have to research it but if you don't clear the route the behavior that I explained is is there the OSPF does care but about the about the internal external and internal route great deal but the routing table doesn't so there is no route preemption but it could be that if they happen to arrive at the same time and clear IP route would be one of those situations when they arrive at the exact same time that the preference could be given to the internal router internal route from the process but as I said at this point it's just wild guessing on my behalf I would actually have to go and this is something I don't know what the answer to you you caught me there and you have it on camera so this is this is a very very good question

Channel: Networking Lessons

Views: 626

Keywords: configrouter.com, Cisco, Cisco Networking Videos, Cisco Networking

Id: SpuzLIpuxcU

Length: 44min 6sec (2646 seconds)

Published: Thu Oct 12 2017