Today we will begin the study of the EIGRP protocol, which, along with the study of OSPF, is the most important topic of the CCNA course.
We'll return to Section 2.5 later, and now immediately after Section 2.4, we'll move on to Section 2.6 "Configuring, Verifying, and Troubleshooting EIGRP over IPv4 (Excludes Authentication, Filtering, Manual Summarization, Redistribution, and Stub Configuration)".
Today we will have an introductory lesson in which I will introduce you to the concept of EIGRP, an advanced internal gateway routing protocol, and in the next two lessons we will look at configuring and troubleshooting the robots of this protocol. But first I want to tell you the following.
Over the past few lessons, we've been learning about OSPF. Now I want you to remember that when we studied the RIP protocol many months ago, we talked about Loops and technologies that prevent traffic from looping. How can you prevent routing loops when using OSPF? Can methods like Route Poison or Split Horizon be used for this? These are questions that you must answer yourself. You can use other thematic resources, but find the answers to these questions. I want you to learn how to find answers on your own by working with different sources, and please leave your comments below this video so I can see how many of my students have completed this task.
What is EIGRP? It is a hybrid routing protocol that combines the useful features of both a distance vector protocol such as RIP and a link state protocol such as OSPF.
EIGRP is a Cisco proprietary protocol that was released to the public in 2013. From the link state tracking protocol, he took the neighbor establishment algorithm, unlike RIP, which does not create neighbors. RIP also exchanges routing tables with other participants in the protocol, but OSPF establishes a neighborhood before starting such an exchange. EIGRP works in the same way.
The RIP protocol performs periodic updates of the complete routing table every 30 seconds and broadcasts information about all interfaces and all routes to all its neighbors. EIGRP does not perform a periodic full update of the information, instead using the concept of sending Hello messages, as OSPF does. It sends Hello every few seconds to make sure the neighbor is still "alive".
Unlike distance vector protocol, which learns the entire topology of the network before making a decision on how to generate a route, EIGRP, like RIP, creates routes based on hearsay. When I say rumors, I mean that when a neighbor reports something, EIGRP unconditionally agrees with it. For example, if a neighbor says they know how to reach 10.1.1.2, EIGRP believes them without asking, “How did you know that? Tell me about the topology of the whole network!”.
Until 2013, if you only used Cisco infrastructure, you could use EIGRP, since this protocol was created back in 1994. However, many companies, even using Cisco equipment, did not want to work with this puncture. In my opinion, EIGRP is the best dynamic routing protocol today because it's much easier to use, but people still prefer OSPF. I think this is due to the fact that they do not want to "bind" themselves to Cisco products. But Cisco has made this protocol public because it supports third party network equipment like Juniper, and if you team up with a company that doesn't use Cisco equipment, you won't have any problems.
Let's take a short digression into the history of network protocols.
The RIPv1 protocol, which appeared in the 1980s, had a number of limitations, such as a maximum number of hops of 16, and therefore could not provide routing in large networks. A little later, they developed an internal gateway routing protocol, IGRP, which was much better than RIP. However, it was more of a distance vector protocol than a channel state protocol. In the late 80s, an open standard appeared - the OSPFv2 link state protocol for IPv4.
In the early 90s, Cisco decided that IGRP needed to be improved and released an improved internal gateway routing protocol, EIGRP. It was much more efficient than OSPF because it combined features of both RIP and OSPF. When we begin to study it, you will see that configuring EIGRP is much easier than OSPF. Cisco tried to make a protocol that would provide the fastest possible network convergence.
In the late 90s, an updated classless version of the RIPv2 protocol was released. In the 2000s, the third version of OSPF appeared, RIPng and EIGRPv6, which supported the IPv6 protocol. The world is slowly moving towards full transition to IPv6, and the designers of routing protocols want to be ready for this.
If you remember, we studied that when choosing the optimal route, RIP, as a distance vector protocol, is guided by only one criterion - the minimum number of hops, or the minimum distance to the destination interface. So, router R1 will choose a direct route to router R3, despite the fact that the speed on this route is 64 kbps - several times less than the speed on the route R1-R2-R3, equal to 1544 kbps. The RIP protocol will consider a slow one hop route to be optimal rather than a fast 2 hop route.
OSPF will examine the entire network topology and decide to use the route through router R3 as the faster one to communicate with router R2. RIP uses the number of hops as a metric, while OSPF uses cost as a metric, which in most cases is proportional to the bandwidth of the link.
EIGRP also focuses on route cost, but its metric is much more complex than OSPF and relies on many factors, including Bandwidth, Delay, Reliability, Loading, and Maximum Packet MTU. For example, if a node is loaded more than others, EIGRP will analyze the load on the entire route and select another node with less load.
In the CCNA course, we will take into account only such metric formation factors as Bandwidth and Delay, they will be used by the metric formula.
Distance vector protocol RIP uses two concepts: distance and direction. If we have 3 routers, and one of them is connected to the network 20.0.0.0, then the choice will be made by distance - these are hops, in this case 1 hop, and by direction, that is, which path - upper or lower - to send traffic .
In addition, RIP uses a periodic update of information, sending a complete routing table throughout the network every 30 seconds. This update does 2 things. The first is actually updating the routing table, the second is checking the viability of a neighbor. If the device does not receive a response table update or new route information from the neighbor within 30 seconds, it understands that the route to the neighbor can no longer be used. The router sends an update every 30 seconds to see if the neighbor is still alive and if the route is still valid.
As I said, Split Horizon technology is used to prevent route loops. This means that the update is not sent back to the interface it came from. The second technology to prevent loops is Route Poison. If the connection to the 20.0.0.0 network shown in the picture is interrupted, the router to which it was connected sends a “poisoned route” to the neighbors, in which it reports that this network is now available in 16 hops, that is, it is practically unreachable. This is how the RIP protocol works.
How does EIGRP work? If you remember from the lessons about OSPF, this protocol performs three functions: establishes a neighborhood, updates the LSDB with the help of LSA in accordance with changes in the network topology, and builds a routing table. Establishing a neighborhood is a rather complicated procedure that uses many parameters. For example, checking and changing a 2WAY connection - some connections remain in the two-way communication state, some go into the FULL state. Unlike OSPF, this does not happen in the EIGRP protocol - it checks only 4 parameters.
Like OSPF, this protocol sends a Hello message every 10 seconds containing 4 parameters. The first is the authentication criteria, if it has been previously configured. In this case, all devices with which the neighborhood is established must have the same authentication parameters.
The second parameter is used to check whether the devices belong to the same autonomous system, that is, to establish an EIGRP neighborhood, both devices must have the same autonomous system number. The third parameter is used to verify that Hello messages are sent from the same Source IP address.
The fourth parameter is used to check if variable coefficients K-Values match. The EIRGP protocol uses 5 such coefficients from K1 to K5. If you remember, when K=0, the parameters are ignored, but if K=1, then the parameters are used in the formula for calculating the metric. Thus, the values of K1-5 for different devices must match. In the CCNA course, we will assume the default values of these coefficients: K1 and K3 are equal to 1, and K2, K4 and K5 are equal to 0.
So, if these 4 parameters match, EIGRP establishes a neighbor relationship, and the devices enter each other into the neighbor table. Next, changes are made to the topology table.
All Hello messages are sent to the multicast IP address 224.0.0.10, and updates, depending on the configuration, are sent to the unicast addresses of the neighbors or to the multicast address. This update does not come over UDP or TCP, but uses another protocol called RTP, the Reliable Transport Protocol, or Reliable Transport Protocol. This protocol checks to see if the neighbor has received an update, and as its name suggests, its key function is to ensure communication reliability. If the update has not reached the neighbor, the transmission will be repeated until the neighbor receives it. OSPF does not have a mechanism to check the recipient device, so the system does not know if neighboring devices have received updates or not.
If you remember, RIP sends out an update of the complete network topology every 30 seconds. EIGRP only does this if there is a new device on the network or there has been some change. If the subnet topology has changed, the protocol will send out an update, but not of the complete topology table, but only the entries with the change. If a subnet has changed, only its topology will be updated. It looks like a partial update that happens when needed.
As you know, OSPF sends out an LSA every 30 minutes, regardless of whether there has been any change in the network. EIGRP will not send out any updates for a long period of time until something changes in the network. Therefore, EIGRP is much more efficient than OSPF.
After the routers have exchanged update packages, the third stage begins - the formation of a routing table based on the metric, which is calculated according to the formula shown in the figure. It calculates the cost and, depending on this cost, makes a decision.
Suppose that R1 sent Hello to R2, and R1 sent Hello to R2. If all parameters match, the routers create a table of neighbors. In this table, R1 enters an entry about the router R1, and R2 creates an entry about R1. After that, R10.1.1.0 router sends the update to the 24/90 network connected to it. In the routing table, this looks like information about the IP address of the network, the interface of the router that provides communication with it, and the cost of the route through this interface. If you remember, the EIGRP cost is XNUMX, and then the Distance value is indicated, which we will talk about later.
The complete metric formula looks much more complicated, since it includes the values of the K coefficients and various transformations. The full form of the formula is given on the Cisco website, but if you substitute the default values of the coefficients, then it will be converted to a simpler form - the metric will be equal to (bandwidth + Delay) * 256.
We will use just such a simplified form of the formula to calculate the metric, where the bandwidth in kilobits is 107 divided by the smallest bandwidth of all interfaces leading to the least-bandwidth destination network, and the cumulative-delay is the total delay in tens of microseconds on all interfaces leading to the destination network.
When learning EIGRP, we need to understand four definitions: Feasible Distance (possible distance), Reported Distance (advertised distance), Successor (neighbor router with the lowest path cost to the destination network) and Feasible Successor (backup neighbor router). To understand what they mean, consider the following network topology.
Let's start by creating a routing table R1 in order to choose the best route to the 10.1.1.0/24 network. Each device is shown with bandwidth in kbps and latency in ms. We use 100Mbps or 1000000Kbps GigabitEthernet interfaces, 100000Kbps FastEthernet interfaces, 10000Kbps Ethernet and 1544Kbps serial interface. These values can be found by viewing the characteristics of the corresponding physical interfaces in the router settings.
The default Serial bandwidth is 1544 kbps, and even if you have a 64 kbps line, the throughput will still be 1544 kbps. Therefore, as a network administrator, you need to make sure that you are using the correct bandwidth. For a specific interface, you can set it with the bandwidth command, and with the delay command, you can change the default delay value. You don't have to worry about the default bandwidth values for GigabitEthernet or Ethernet interfaces, but be careful when choosing a line speed if you're using a Serial interface.
Please note that in this diagram, the delay is supposedly indicated in milliseconds ms, but in reality it is microseconds, I just do not have the letter μ for the correct designation for microseconds μs.
Pay close attention to the following circumstance. If you issue the show interface g0/0 command, the system displays the latency in tens of microseconds instead of just microseconds.
We will look at this issue in detail in the next video on configuring EIGRP, for now, remember that when you substitute the delay values \u100b\u10bin the formula, XNUMX μs from the circuit turns into XNUMX, since the formula uses tens of microseconds, not units.
In the diagram, I will mark the interfaces with red dots, which include the shown throughputs and delays.
First of all, we need to determine the possible Feasible Distance. This is the FD metric, which is calculated by a formula. For the section from R5 to the external network, we need to divide 107 by 106, as a result we get 10. Next, we need to add a delay equal to 1 to this bandwidth value, because we have 10 microseconds, that is, one dozen. The resulting value of 11 must be multiplied by 256, that is, the value of the metric will be 2816. This is the value of FD for this section of the network.
Router R5 will send this value to router R2, and for R2 it will become the declared Reported Distance, that is, the value that the neighbor told him. Thus, the declared distance RD for all other devices will be equal to the possible distance FD of the device that reported it to you.
Router R2 performs FD calculations on its own data, that is, divides 107 by 105 and gets 100. Then it adds to this value the sum of the delays on the route to the external network: R5 delay equal to one ten microseconds, and its own delay equal to ten tens. The total delay will be 11 tens of microseconds. We add it to the resulting hundred and get 111, multiply this value by 256 and get the value FD=28416. Router R3 does the same, receiving the value FD=281856 after the calculations. Router R4 calculates the value FD=3072 and sends it to R1 as RD.
Please note that router R1, when calculating FD, does not substitute its bandwidth of 1000000 kbps into the formula, but the lower bandwidth of router R2, which is equal to 100000 kbps, because the minimum bandwidth of the interface leading to the destination network is always used in the formula . In this case, routers R10.1.1.0 and R24 are located on the way to network 2/5, but since the bandwidth of the fifth router is larger, the smallest bandwidth value of router R2 is substituted into the formula. The total delay on the path R1-R2-R5 is 1+10+1 (tens) = 12, the normalized throughput is 100, and the sum of these numbers, multiplied by 256, will give the value FD=30976.
So, all devices have calculated the FD of their interfaces, and router R1 has 3 routes leading to the destination network. These are routes R1-R2, R1-R3 and R1-R4. The router selects the minimum value of the possible distance FD, which is 30976 - this is the route to the router R2. This router becomes the Successor, or "successor". The routing table also indicates Feasible Successor (backup successor) - it means that in the event of a break in communication between R1 and Successor, the route will be carried out through the backup Feasible Successor router.
Feasible Successors are assigned according to a single rule: the advertised RD of this router must be less than the FD of the router on the leg to Successor. In our case, R1-R2 has FD=30976, the RD in the R1-K3 section is 281856, and the RD in the R1-R4 section is 3072. Since 3072 < 30976, R4 is selected as the Feasible Successors.
This means that in case of a communication failure in the R1-R2 network section, traffic to the 10.1.1.0/24 network will be routed along the R1-R4-R5 route. Route switching when using RIP takes several tens of seconds, when using OSPF - a few seconds, and in EIGRP it happens instantly. This is another advantage of EIGRP over other routing protocols.
What happens if both Successor and Feasible Successor lose their connection at the same time? In this case, EIGRP uses the DUAL algorithm, which can calculate a backup route through a likely successor. This may take a few seconds, during which time EIGRP will find another neighbor that can be used to carry traffic and put its data into the routing table. After that, the protocol will continue with the normal work of providing routing.
Thank you for staying with us. Do you like our articles? Want to see more interesting content? Support us by placing an order or recommending to friends, 30% discount for Habr users on a unique analogue of entry-level servers, which was invented by us for you: (available with RAID1 and RAID10, up to 24 cores and up to 40GB DDR4).
Dell R730xd 2 times cheaper? Only here in the Netherlands! Dell R420 - 2x E5-2430 2.2Ghz 6C 128GB DDR3 2x960GB SSD 1Gbps 100TB - from $99! Read about
Source: habr.com