OSPF Metric –Cost

5. Manual Cost

And those other options are, like, we can go with the second method. So now this is like a second method where we can go and manually change the cost directly on the interface irrespective of what the bandwidth is. So in the second option, what we'll be doing is, I'm saying that the cost is going to becalculated isolated or it is directly configured under the interface, which is going to influence the best path again. So like in the previous, it is going to automatically reference the bandwidth to the power of eight divided by bandwidth, which is the default, and if you want to influence the best route, you just go and change the interface bandwidth, which is again limited to only 100 MPs cost values. So in the second option, what I'm doing is directly going to the interface and we have a command called Iposp of cost and I can change the cost values directly on the interfaces. So this is equivalent to a manual cost default. But you can see I'm going to these two interfaces. When I change the cost value, it is going to show me the cost values, and again, the best route is the sum of all the costs to that particular destination. Like in my example, you can see here, I made some changes I think. So let me see what the changes are that I made. I went to this interface.

I'll just remove what I did in the previous I think on the Gig interface. I have changed the cost from our previous topic, so I'm going to remove that. I think I added that. Let me remove that one if you verify the default cost values, so the default cost value for the kicking tenant will be one and the default cost on the serial link will be 64 based on ten to four of eight divided by bandwidth. Now in my case, what I want is a serial interface. The cost should be manual 64. I'm simply going and saying that the cost of this particular interface, let's say five. So, if you verify the IP OSPF interface you can use, you will notice that the cost value has changed to five. Now based on this, the best route calculation will also change, which means if I go and check out OSPFnow, what you should see is that the cost on this interface is five, and this is the default 64 one one.So from the router one to reach the two networks, the overall cost is six, and that's what you can see here. To reach this one, the network is 70.70, which means I changed the 564 plus 164 plus 569 plus 170, as you can see. So this is another way where we can directly go onto the interface and decide our own cost values. As for this example, you can see the cost values are based on the default tender powerdivided by bandwidth based on the default cost of any link above or equal to 100 MPs. Another interface that we have changed.

6. Auto-cost Reference Bandwidth

And the next option is that we can use something called OSP of metrics. We can influence the OSP cost based on the third option, that is auto reference bandwidth. The default reference will be tended to offer divided by bandwidth, which means it's going to use 100 MPs. The default reference will be divided by whatever bandwidth is on the interface. So whatever the bandwidth of the interface is based on this. So we discussed the disadvantage of this one, which is that regardless of the speed you use, say 100 Mps or one gig, ten gigs, or even 100 gigs, the cost values will be derived as one only. So that is the default behavior. So what I want is to change this default reference to some other value. I want to scale for high-speed links. So what I can do is I can go on to the router and I can change the auto cost reference bandwidth option. So with this option, what I'm doing is the default reference, which is taken as 100 MPs divided by bandwidth. So I'm going to change its value. So if I'm changing the value as autoreference bandwidth, instead of setting it to 100 MPs, I'm going to set it to something like 100 gig. Like here, this is actually 100 gigabytes. Now in this scenario, the link will scale, which means I'm going to set it as this and instead of 100, I'm just changing it to 100 gig. So when you change this automatically, now depending upon the calculation, if you're using a 100 gig link, obviously the cost will come to one, and if that particular link is 40 gig, then the cost value will come to four, depending upon this calculation. Again, you just need to calculate these MVPs. You just write down that and 40 gig. You write down that. Basically, the calculation is something we don't need to really worry about because at the back end, the routers will do this process. But when you change the reference bandwidth to this value, these are the possible cost values we get. For all 100 MPs, 100 gig links, the cost will become one, and for 40 gig links, the cost will be four, and for ten gig links, the cost will be ten. And of course, for the one gigalink, the cost will be 100. Based on this, we can change, like let's say I'm going to the router and I'll try to make some changes here, okay? So as for the default values, what I discussed, let's go and quickly verify this. So what I can do is that the default reference will be 100 MPs. So in this scenario, I want to change the default reference to 100 gig. So probably 100 Gig means we need to say 100 and then probably three zeros.

And again, when you are verifying this, make sure that you're verifying it on the packages. Packages are set. I verified that it's not showing me the exact values. So I've put this in my GNS three topology. So what I did is if you just go and verify,I'm saying route OSPF one, let me first remove the configuration. If you go and verify share IP OSPFinterface brief, you can see the default values like theserial link has a cost of 64 and the fastest link has a cost of one based on 100 MPs. So I'm going to change the reference to ten G. So how to change auto cost reference bandwidth? I'm going to say probably 100 gig. Again, this value we give you in terms of MVPs, which means if I'm giving this value,it's like 100 kg and once you change the values,it is a show iPOST of Interface Brief. So our verification is on this fast Ethernet link. So as per the fast internet link, normally any link with the fare, the internet, of course, you don't have that here. The cost value will go to 1000, and that's what you can see here, 1000. So, similarly, if the link speed is one gig, in that case, the cost value will be 100 gig.

Okay? So again, Cisco recommends using these OSPF reference bandwidth settings on all the routers again in the enterprise network. So if you're using the first option, like I discussed, there are three options we discussed previously. So in the first option, again, the problem is that it is going to consider the interface bandwidth, but based on the tens of port opera divided by the bandwidth option. So this means up to 100 MPS link. This is good, but in your production network if you're using anythingabove 100 MPS links, then again, either you change the reference bandwidth, so by default it takes 100 MPS, you can change this to 200 gig also, as we have seen just now, or you can go and manually set the costs on each and every interface. So again, the cost values in the first and third cases, like in this example, are going to be considered the interface band. The only difference is that the reference values will change depending on that. So here it is, ten to the power of eight divided by bandwidth, the default, or you can change it to 100 Kg, like we did just now. Or you can simply go and set the price. When you are setting the cost,it's like the manual cost.

OSPF Areas

1. OSPF Single area – Lijmitations

So the next concept will move on with the concept of OSPF areas. Like previously, we have discussed the areas, probably in a single area. So in this section we'll try to understand why we need areas, what the limitations of using a single area are, and why there is a need to divide a bigger topology into multiple areas. So let's get started. Here you can see OSPF by default. In our previous lab, what we did is we connected three routers and then we configured all the routers in one single area, just like one group. But again, this one area is like placing all the routers in just one area or in just one database, just like one database. But most of the large Osuf networks basically have some drawbacks or limitations with this design. Like, especially if you take an example, we have some 900 routers, like, let's say you have 900 routers connected, probably to different sites. And so you probably have thousands of routers. So, let's say you have someplus interfaces on each and every network, such as one LAN and two Van connections, or even more. So each and every interface will probably have one network. So if you just think about the number of networks or the size of the routing table, or the size of the database table, generally you just think about it, it's very large. Again, so especially the normal OSP of what we have done up to now based on a single area. So that is not going to be designed, that's not going to be scaled for bigger design, especially now. The main reason is that it takes a lot of CPU resources to run the SA algorithms inside the database, and also it has more impact on the convergence time. Also, it will be very slow in reacting to changes.

Let's see more details, like what are the different problems we see here? One of the issues is that we discussed how the OSPF will operate. So the algorithm is going to use some kind of LSS,the link state advertisements, where every router is going to advertise its own information in the form of LSS to every other router or its neighbors, and that router is going to forward that to the next router, and then that router is going to forward it to the next router. In a similar way, if you assume you have 900 routers and every router is advertising their own information in the form of LSS, which means every router issending out the arrangements to remain almost like 900 or 899 routers, then you can see the utilisation of the routers will be very high. So it takes a lot of CPU resources to advertise and build a database that links the databases, and that's probably going to increase the resources in general, the memory and the CPU resources. And again, it takes a lot of CPU to run the algorithm and calculate the best route because it has to calculate all the possible routes. That's one issue. So basically, the problem comes with most of the small routers. In particular, you will be connecting some high-end routers here, the core routers, and probably on the endlocations you might be running some small routers, let's say 1900 or 800 series, and this router will be advertising or receiving the advancements for all the 900plus routes of the routers.

As a result, this router's stabilisation will most likely be very high, and it may not be able to handle that many requests. Of course, the routing table will be larger as a result of the large number of routes you have. So this router has to maintain the database and also find all the best routes to reach that particular and every router inside the database. So most of the problems come with the low-end routers because they don't have enough memory to handle such a huge database. So again, that's one problem, and the next problem is too many eradasmen Because your network is bigger than 900plus routers, every router is sending an advancement. So basically, I'm receiving almost more than 900 plus arbitrage from each router and I'm going to maintain that in my links database. Again, that requires more CPU and memory resources, of course, to meet the processing requirements. And once I built that, again, let's say there is some change. There is a new route added or a link goes down. In that scenario, what happens? So, if the status of any interface in your network changes, Whether it goes up and down and forces every router to run the SPF algorithm again depends on whether they received a new LSA from this router. Again, they have to recalculate the algorithm and maybe based on that, they have to figure out the shortest paths. Calculate the cost. That will again increase the CPUT utilisation again.So, the bigger the size of the network, the more resources it will utilize.

So that is one limitation. So the number of advertisements required is more, so more times it will receive the advancements. Of course, it has to run the algorithm too many times. So these are the general issues that you generally face when you have OSPF on a large network. Again, at the same time, as I said, the convergence time, like if this fails, then they do advisementsLSS and probably the router has to figure out the alternate path, and that will also take time because the OSPF has to arrange the and then calculate the rhythm, which will again impact the convergence time as well. So these are general issues. Like I said, the issues Here, it takes more CPU resources to run those algorithms and also the convergence time because it's a little bit slower to react to changes in the network. And the bigger the network, the more convergence time it may take as well as more CPU resources. Sometimes if you have this router, like an 800 or 900 router, and if it is unable to process them, basically that can restart, that can go down, and that will impact the communication also.

2. OSPF Multiple Areas

Okay, so the next thing we'll try to understand is the concept of areas here, how they impact theindigenous networks, how it's going to be. Like we have seen in the previous section, there are some limitations with single areas, especially when you're using plenty of routers, hundreds of thousands of processes in a single area. The problems are that the SPF algorithm may run too many times, you may receive too many advancements. And also, the OSPF database table, the links to database table, as well as the routing tables may be bigger depending upon the size of the network. So to overcome this, we have a solution. And the solution is like this: areas. So in areas, what we'll do is logically group a set of routers in one area.

So this is just like we have a large database. You've got some thousands of routers in your network and you've got a database for thousands of routers. So what you're doing is you're just dividing them into small, small groups. Just like you can say, let's say if you take an example of a simple classroom, let's say I've got some 200 students, I'm going to manage them better by making them into different sections. So I'm going to divide them into four different sections, ABCD, and I'm going to have 50 students in one group and separate them into something like this. So, instead of making or placing them all in one area or group, we use the same concept here. So areas like logical grouping, so you can place everyone in one area or one group, or you can group them in different groups. I'm going to use different areas here. So this makes the size of the database smaller. Because using the same example as I said, let's say you've got some network with 100 routers and 200 subnets. We are going to divide them 100 into ten groups, which means each area will be ten like that. So each group will have ten routers. So basically, this will minimise the size of the database. Like here you can see the groups, the grouping, numbering,numbering. We do any number, we give the area number, like I'm giving area ten here, area twenty here,area thirty here, and area zero here. Again, we'll talk about the rules in the next few topics. So let's see what the benefit of this is. So the main benefit is that it minimises the CPU and memory resources. Because now this router is going to maintain the database within the same area. And whenever this router sends an advertisement, the linksto advertisements, these advertisements will be sent only to the routers within the same area, and only these area routers will participate in the algorithm, and only these area routers will maintain the database.

Which means the database maintained by this area will be different from the database maintained by this area. and the database maintained by this area. So every area will have a separate database which includes the advertisements which are generated within that specific area. So this will result in a smaller database because the database is restricted to that particular area only. So this will minimise the CPU as well as the memory resources of individual routers. And the next thing is any changes that occur, let's say when this router sends out an advanced LSS, whether for the first time or whenever there is any change that occurs. So these LSS, or the links for advancement, go in this particular area only. So, like, if you have hundreds of neighbours and I'm just grouping 30 here. 30 years. 20 years or 30 years. 30 years, let's say. So the advertisements will go only to the routers within that particular area. So only to those particulars, the flooding of the rudiments will be restricted within that particular area and if any changes occur, So basically, the advancement has to be sent to only the study routers. As a result, if any changes occur, you will experience faster convergence. So between the areas they do exchange, again, they do exchange the best routes, but again, advertisements are restricted to that particular area only. As you can see here, for example, if you have 100 routers, each router has two networks.

So now when you create ten areas, you get approximately ten different routers into ten areas again, so the speed of calculation on the router will have to process only the topology on the ten routers. So the calculation, the redirect means are not processed to the entire domain of the entire network, but rather to that specific area only, rather than on all of the routers, all of the 200 links here. So again, the same points as we discussed: what are the advantages of using multiple areas? The same area is a logical grouping of the routers. It minimizes the CPU and the memory resources because it will automatically minimise the processing overhead on the routers because all the routers within the same area maintain the same database. So the database of area one will be different, the database of area zero will be different, and the database of area two will be different. So there is no need to maintain the database of each and every router in your network, so you can restrict that to that particular area only. Again, as I said, any changes are dashed within the area only, so it is being flooded. So if any new links are added, So the advancements will be within this particular area only. restricted to that particular area only. So any changes are going to impact only the routers within that area, and with the help of this, it will provide you with faster convergence because when you have a lower number of routers to receive advancements, it will result in faster convergence because restricting any changes within the area means advancements will not go outside the area. And when it is calculating the best route, the algorithm is applied only to the changes within that particular area only. So it's like restricting.

Study with ExamSnap to prepare for Cisco CCIE Service Provider Practice Test Questions and Answers, Study Guide, and a comprehensive Video Training Course. Powered by the popular VCE format, Cisco CCIE Service Provider Certification Exam Dumps compiled by the industry experts to make sure that you get verified answers. Our Product team ensures that our exams provide Cisco CCIE Service Provider Practice Test Questions & Exam Dumps that are up-to-date.