CC BY
37
ANALYSIS OF THE EFFECTIVENESS OF THE EIGRP ROUTING PROTOCOL
Olga A. Abramkina,
Siberian State University of Telecommunications
and Information Sciences, Novosibirsk, Russia, Keywords: EIGRP protocol, lvalue,
[email protected] path, metrics, constraints.
Considered an improved remote-vector protocol EIGRP for selecting the path in the context of changing the k-values of the weight metric. The weighted metric of the EIGRP protocol has 5 values: bandwidth (Kl), load (K2), delay (K3), reliability (K4) and MTU (K5). In the Cisco specification for calculating the weighting metric are used only the bandwidth and delay coefficient, Kl and K3, respectively, with a value of 1, the others are set to a value of 0. In this paper, a study was made of the operation of the network with a change in the values of the default coefficients, since they can vary from 0 to 255. A network analysis has also been performed, including the coefficient K2 in the weighted metric with the value from 0 to 2.
It is proved that when using a metric with different coefficient variations, some useful information is lost and a situation can occur when the correctly found path does not satisfy the constraints. As a result of the protocol analysis, an attempt was made to select the optimal coefficient values for the selected network. To achieve this goal, VMware version 12.5.2 and GNS3 version 2.0.0b2 were used as a server as a network simulator. The network under investigation is based on routers of the c7200 model with serial connections. The results of modeling the network using the EIGRP protocol under the conditions of changing the weight coefficients showed that the use of the default coefficient values is advisable in the case when there are no stringent network requirements or the shortest path search. If load balancing requirements occur both within the same network and between networks, you can change the values of the coefficients and increase their number (include, for example, K2) but no more than one from the default values. Changes more than one can lead to network overloading during balancing, as well as the impossibility of determining the shortest path.
Information about author:
Olga A. Abramkina, Siberian State University Of Telecommunications And Information Sciences, Novosibirsk, aspirant, Russia
Для цитирования:
Абрамкина О.А. Анализ эффективности работы протокола маршрутизации EIGRP // T-Comm: Телекоммуникации и транспорт. 2017. Том 11. №10. С. 60-65.
For citation:
Abramkina О.А. (2017). Analysis of the effectiveness of the EIGRP routing protocol. T-Comm, vol. 11, no.10, рр. 60-65.
T-Comm Том 11. #10-2017
Introduction
One of Hie tasks of Successful traffic transmission over the network is the organization of the path. Today several protocols are widely used, such as OSPF, BGP, El GRP. RIP, IS-ES. Some of them use the remote-vector principle when choosing a path, some of them are based on the values of the state of links. Whatever the principle, it is necessary that the protocol ensure the timely search for the best path. It is best to consider such a path, for example, in which there arc no loops that has sufficient capacity to transmit the required resource, is the «shortest» {compared to all available at the time of transmission) and is sufficiently reliable, etc. Often, it is impossible to ensure that all these criteria are fulfilled in a single protocol. The implementation will be quite complex and may not justify the expectations placed on it. Since the network will play by other criteria, for example, increasing reliability, the number of possible route variants under force majeure on the network will decrease.
Thus, it is advisable to select protocols for a specific network task. Next, examine the effect of changing certain parameters on the network and identify the best values.
The article discusses the dynamic routing protocol EIGRP, which within seconds can learn about network problems and rebuild its routing tables. In this case, the packages will be sent already on the current path.
Another important point is the balancing of traffic. Dynamic routing protocols support this function and you do not need to add redundant routes manually by calculating them.
The introduction of dynamic routing greatly facilitates the scaling of the network. When adding a new element to a network or subnet on an existing router, you need to take several actions to make everything work, and the probability of error is minimal. At the same time information about changes instantly diverges across all devices. Exactly the same can be said about global changes in topology.
1. Features of the EIGRP protocol
EIGRP is a protocol for internal gateway routing, using the distance vector principle. Consists of four main components:
1. Neighbor Discovery / Recovery.
2. Reliable Transport Protocol/
3. The block of final states of algorithm DUAL (DUAL Finite State Machine).
4. Protocol Dependent Modules.
The EIGRP protocol is an improved version of the IGRP protocol, which is independent of network layer protocols.
The Detection / Recovery, process allows you to recognize other rotiters on the networks to which they are directly connected, and also recognize the lack of access to the neighbor or the termination of its operation. This process is provided by sending small hello packets of type «Hello», with unproductive costs very small. While the router receives «Hello» packets, it can determine that its neighbor is functioning normally. Once this is determined, the neighbor can exchange information routed.
One of the features of the EIGRP protocol is that it has a reliable RTP delivery protocol in its structure that supports a multicasting mode and allows you to regulate the need to send EIGRP packets to all neighbors. This approach also reduces the convergence time, especially in communication channels with different speeds. A reliable transport protocol supports multi-type trans-
mission of packets in both multi-send mode and single-sending mode. Some EIGRP packages must he transmitted with a high degree of reliability, for others this condition is not necessary. To increase efficiency, reliability is provided only if necessary. For example, in a network with multi-access and multi-sending capabilities, such as Ethernet, there is no need to send trust-enhancing «Hello» packets to all neighbors individually. Therefore, EIGRP sends in a multi-send mode one «Hello» packet with the indication (written in the packet) informing the recipients that the reception of this packet does not need to be acknowledged. Other types of packages, such as «Update», require acknowledgment of receipt, which is indicated in the package. A reliable transport protocol is provided by means of fast packet transmission in multicasting mode in the event that unacknowledged packets wait sending. Such tools help not to increase the convergence time in the presence of communication channels operating at different speeds.
Updatable algorithm DUAL, which allows obtaining independent cycles from each time of route, calculation w hile having support of other protocol groups. Regardless of the changes in the structure of the EIGRP protocol, its basic remote information is not changed, so the networks generated by the IGRP protocol can easily be converted. And the time of such convergence is less, in comparison with other routing protocols. The finite slate block of the DUAL algorithm implements the decision-making process for calculating all routes. The block tracks all paths declared by all neighbors. Remote information is an indicator that is used by the DUAL algorithm to select effective paths that do not contain loops. The DUAL algorithm selects parhs that arc included in the routing table based on the principle of probable subsequent elements. The next element is a neighboring router used for packet transfer and having the cheapest route to the destination, with the guarantee that such a path is not part of the routing cycle. When there are no probable subsequent elements, but there arc neighbors announcing the destination, it is necessary to recalculate. A new follow-up element is defined. The conversion time affects the total convergence time. And although recalculation does not require intensive CPU usage, try to avoid them unnecessarily. When the topology is changed, the DUAL algorithm checks for the presence of probable subsequent elements. If they are present, the algorithm uses everything that it detects to prevent unnecessary recalculations. In more detail, the possible subsequent elements will be considered below.
As mentioned earlier, the EIGRP protocol supports groups of other protocols, and special modules, which in turn depend on network layer protocols, help it. For example, the IP-17.IGRP module is responsible for sending and receiving EIGRP packets encapsulated in the IP protocol. The IP-EIGRP module is responsible for analyzing (decomposing) EIGRP packages and notifying the DUAL algorithm of obtaining new information. IP-EIGRP refers to the DUAL algorithm for making routing decisions, the results of which are stored in the IP routing table. IP-EIGRP is responsible for redistributing routes detected by other IP routing protocols.
Based on the features of the EIGRP protocol, it follows that load balancing can be performed between paths that have the same metrics as well as the metrics of each of them. To determine the total metric, it is necessary to use the following formula
(i ■N
M =
Kt-Bw+*yBW .^.py-**
256 - Load
Ka+R
■256'
(I)
Where BW is the bandwidth; D - delay; Load - load the channel; R - reliability of the channel; Kl, K2, K3, K4, K5 are weights.
The capacity metric (1,544 kilobits) is a static value for calculating the routing metric. Bandwidth is displayed on the screen in kilobits. Most serial interfaces use a default bandwidth value Of 1544 kilobits or 1,544,000 bits/s {1.544 Mbps). This is the throughput of the T1 connection. Some serial interfaces use a different default throughput value. To check the throughput, use the show interfaee command.
The bandwidth value may or may not reflect the actual physical throughput of the interface. Changing the bandwidth does not change the actual communication bandwidth. If the actual communication bandwidth is different from the default throughput, you should change the bandwidth value.
The delay is calculated as the sum of tens of microseconds of all outgoing interfaces to the destination network. To understand what the number is, you need to look at the outgoing interface (show interface gigabitethemet 0/0), 11 nd delay and divide the value by 10. This is the number that will be used for the calculation. The total delay is calculated by multiplying by 256.
Reliability of the communication channel is calculated as the value of the frames received with ail error to the total number of received frames on this interface. The minimum value is chosen all the way to the destination and can take values from 255 to 0. 255 is 100% reliability of the channel, 230 is 90% reliability, etc. In modern networks is practically not used. Although this value may change dynamically, it is not a trigger for sending updates (KIGRP Update).
The load is calculated as the ratio of the current channel utilization in the outgoing direction (Tx load) to the bandwidth. It is selected as the maximum value on the entire path e to the destination, Similarly, like reliability, despite a regular dynamic change, it is not a trigger for sending an update. The maximum value of 255 corresponds to the fully loaded channel, the minimum value of 1 is a completely empty channel.
The minimum size of the MTU is all the way to the destination, It is supposed to avoid the fragmentation of IP packets if transmitted through interfaces with a maximum MTU of 1500 bytes. However, Lhis parameter has no eflees on the calculation of the metric and is not taken into account in its spreading. Even though this value is passed to EIGRP Update, it is never used to select the best route. The MTU indicator is taken into account in quite specific cases; for example, if the router must discard one of the routes equivalent to the other characteristics, he will choose the one with the smaller MTU.
Each route in the EIGRP protocol is characterized by two numbers: Feasible Distance and Advertised Distance (sometimes you can see Reported Distance instead of Advertised Distance, it's the same thing). Each of these numbers is a metric or the cost (the more - the worse) of this route from different measurement points: FD is «from me to the destination» and AD is «from the neighbor who told me about this route to the place Destination».
Each subnet EIGRP knows on each router has a Successor router from among its neighbors, through which the best (with a smaller metric), according to the protocol, the route to this subnet. A subnet can also have one or more alternate routes (the
neighbor router through which such a route goes is called Feasible Successor). EIGRP is the only routing protocol that stores Lhe spare routes (in OSPF they are, but are contained in the topology table and they must be previously processed by the SPF algorithm), which gives it a plus in speed: once the protocol determines that the main route (via successor) is unavailable, He immediately switches to spare. In order for the router to be feasible successor for the route, its AD must be less than the FD successor of this route (that's why we need to know AD). This rule is used to avoid l ings diring routing.
By default, weights have the following values: Kl = 1; K.2 = 0; K3 = I; K4 = 0; K5 = 0. Taking into account this fact, the calculation formula has the form:
M = (KrBfV+KJD)-256
(2)
For such values, the shortest path will be selected by the smallest metric.
The case when the weighting factors are changed and their influence on the network lias not been sufficiently studied to date. To confirm or disprove the default values recommended by Cisco, we examine the network with different coefficients parameters,
N. Implementation of the network based on the EIGRP protocol
To build the network, the GNS3 environment version 2.0.0b2 was used as a network simulator, VMware version 12,5,2 was used as the server. The network consists of five routers of the c7200 model with a connection between them via a serial interface (Fig. 1).
The necessary interfaces on the RI and R5 routers are configured as a loopback, modeling the client's network, and the interface towards the provider. An example of setting on router 1 is shown below:
Reconfigure terminal Rl(config)# interface loopback 1 Rl(config-if)#ip address 10.1.1.1 255.255.255.0 Rl(config-if)# interface s0/0 Rl(config-if)#ip address 199.1.1.1 255.255.255.252 RI (config-if)# no shutdown
Similarly, the interfaces are configured on the other routers. After enabling the interfaces on each router, it is necessary that the client 10.1.1.1 and 10.2.2.2 local networks see each other. To do this, you must configure the network protocol EIGRP [2]. The configuration of the EIGRP protocol is similar to the setting of the OSPF protocol. The router eigrp command allows its use and puts the user in EIGRP configuration mode, where one or more network commands are then configured. For each network interface command, the EIGRP protocol tries to locate neighbors and announce the subnet connected to this interface. Example of configuring the router Rl and R2:
Rl(config)# router eigrp 100 R1 (config-router)# network 199.1.1.0 R1 (config-router)# no auto-summary
R2(eonfig)# router eigrp 100 R2(config-router)# network 199.10.1.0 R2(config-router)# network 199.1.1.0 R2(confîg-router)# network 199.11.1.0 R2(config-router)# no auto-su miliary
Fig, 1. Schema ofihe network under investigation
After setting the EIGRP protocol, weights will be specified by default (Fig. 2).
LlfabSV ip reute
icdea: C - coanttrtesi. 5 - «title, » - HIP, K - ucblle, a - S3?
D - IlCftP, El - LIGSP external, 0 - OSW, 1A - DSP? inter area HI - CSPF «3* «ttmü type î, JÇ - G3ÎT N33* eRfttftfti type J E3 - Q5PF e*Eenwl typ« 1» IZ - CSPF iitetfial tïpe 2 1 - IS-IS, n - ÎS-1S suBRMTi. 11 - IS-IS lml-l, U - IS-IS level-2 u - !MJ later area, * - candidate <1*fault, 0 - per-tieer itatic route 0 - OUR« P - periodic dowelsa-^M static taure
latevav of last resort u not Ht
139.1.1.0/30 13 aataetted, I nifcsecs : 199.1.1.0 is directly esneeeted, feneJ.1/0
199,¡.1.0/35 «tamced, ï ïttttew
> :«.:,t.c na 199.1.1.2, 00:01:0«, stntii/o
199.10.1.0/30 11 abetted, 1 j'aljr.eca 5 149.10.1.0 |*0/JM3BSÉ| via 199.1.1«!« 00:01:49, Serial1/0 199.11.1.0/30 ta Mfceetted, % euh&eia
199*11.1*6 [90/2C3ieS«| via 199.1.1,:, 00i0lM9, Serlall/e 199.11.1.0/30 L> Mfcaetted, 1 »atceis I 199.13.1.0 [M/3193ÎS€| vu 199.1.1.2, 00:01:», îerlall/0
10.0.Q.0/11 it acfeoetted, î ïtib&st» ) a JO.J.Ï.O tnO/3î31S5«l vi» 159.1.1,J, 00:00:33, 5«Ull/0 ' 10.1.1.0 le directly connect«]. Locgfeackl
Fig. 2. The default EIGRP protocol parameters for RI
Once enabled, the EIGRP protocol starts announcing the subnets connected to the interfaces, sending out «Hello» messages and listening to them, trying to establish neighborly relations with other EIGRP routers.
The path metric is one of the most important parts of any routing protocol, as it is a condition for adding a path to the routing table. The more metric parameters are better (the value of the metric is smaller), the greater the probability that the network device will decide to send the packet along this path. Modern routing protocols in calculating metrics try not to exceed the default values and use no more than three metric values.
Among them, the bandwidth, delay and percentage of lost packets. Each of these metrics affects each other. The combination of metrics makes it possible to improve the routing algorithm, These metrics provide only a general estimate of communication channels, and by changing their values, you can get a more detailed estimate and improve efficiency.
If you change any metric in the protocol, you need to configure the value of this metric on each router, otherwise the neighborhood will be disconnected. Neighborhood is automatically configured. To change metrics, delay is usually changed, because bandwidth affects QoS parameters, in addition, bandwidth change does not always change the metric value (if the worst link has not changed).
EIGRP includes regulation to use 50 percent of the installed bandwidth. Reducing bandwidth can cause various problems. For example, because of the reduced bandwidth, EIGRP neighbors may stop receiving complimentary packages. The delay variation does not affect other protocols and does not cause EIGRP to reduce the bandwidth.
The minimum value of the MTU is actually calculated, but assumes no role in determining the best path. The coefficients used to calculate the metric can be declared manually. This is done by the command metric weights tos kl k2 k3 k4 k5 (tos = 0): Bandwidth (Kl), Load (K2), Delay (K3), Reliability (K4), MTU (K5).
Example of adding k2 = 1 to R! (Fig. 3): Reconfigure terminal Rl(config)# router eigrp 100 R1 (config-router)# metric weights 01 1 100
To change the «Hello» and «Hold» timers, you can use the command «ip interval hello eigrp autonomousserverdevice timer» and «ip holdtime eigrp autonomousservernumber timer» in interface configuration mode.
To manually change the route of the routing protocol, you can specify the bandwidth and bandwidth values using the «bandwidth value» and the «delay value» on the interfaces. Also, you can set how many routes with the same metric will be simultaneously used for load balancing using the maximum-paths command, and enable load balancing across channels with different metrics using the variance command.
EIGRP network commands allow two types of syntax: one with a mask pattern at the end and one without it (command network 199.1.4.0), Without a mask pattern, this command must output a class network (class number A, B or C network number). Being commanded, this command tells the mariner to do the following: find their own interfaces with addresses in this class network; Enable the EIGRP protocol on these interfaces.
Example of adding k2 = 2 to Rl. RI ^configure terminal Rl(config)# router eigrp 100 R1 (config-router)# metric weights 0 12 10 0
Figure 2 and Figure 3 show all possible routes that are available in the routing table with the corresponding metrics [D/M], where the route is marked as D (Distance), and the metric is M (Metric).
A route can have three values:
1) Value 5 if the route is total;
2) A value of 90 if the route is internal;
3) The value is 170 if the route is external.
T-Comm Vol.11. #10-2017
7TT
■way of Lut resort is cot net
199.1.1,0/30 13 snbftetted, 1 sobnets
199.1.1,0 if directly connected« Serial1/0 199.2.1.0/10 la autoested, 1 subneta
199.2,1.0 (90/3100357] via 199.1.1.2, OOiOl:«, 5erUll/0 199.10*1,0/30 19 sub&ettcd, 1 subnets
199.10.1.0 ]90/2i&6357] via 199.1.1.2, 00:03:06, Serull/Q 399.11.1.0/30 la subnetted, 1 aubcets
199,11.1,0 190/2688357] via 199.1.1.2, 00i03;0€, Serl*ll/0 199.12.1.0/30 is subletted, 1 subcets
199.12.1.0 150/3200357] via 199.1.1.2, 00:02:19, 5ena.ll/0 10.0.0,0/21 ia subaeccta, I subsets [ 10.2.2.0 1170/3326357] via 199.1.1.2, 00:00:15, Serial1/0 10.1.1.0 19 directly connected, Loopfcacfcl
Fig. 3. The parameters of the E1CRP protocol for k2 = ) for RI
In the investigated network, there are both internal routes and external ones. Let's compare the result of the experimental calculation of the metric with the practical one for the path from 199.2.1.0 to 199.1.1.2 by default. To do this, we calculate the metric by the formula (1) with the following initial data:
BW = 1544 kbit/s,
D = 40000 ms,
K1 =K3 = 1,
K2 = K4 = K5 = 0.
Mf = (l 000000 /1544 + 40000 /10) ■ 256 = 2682031,
Mexp — 2681856.
Mexp< Mp
The total metric, calculated by the router is less than that calculated manually. This indicates a more accurate calculation of the values of the total metric by the router. To investigate the network, this error can be critical, since the EIGRP protocol just chooses the best path for the lowest metric value.
As noted earlier, the weighting factors affect the metric value. A study was conducted in which the coefficient K2 was subjected to changes. Its values ranged from 0 to 2 (0.1.2 respectively) [3]. The choice was made to introduce this coefficient because it is responsible for downloading the channcl in the route. This value is dynamic, it can change and lead to a constant recalculation of the route metric. Nevertheless, the research is aimed at revealing the possibility of choosing such a vaiue of the weight coefficient, so thai this would not be necessary [4].
Based on the results of the network testing (Fig. 1), the dependencies of the packet transmission rate are plotted depending on the route metric (whether it will pass through one hop or several) for different values of the load factor (Fig. 4).
-Ü-iiJUOCT; ': -:
-C-Kähopih
-Ü-iLJnsp» K.'fi Trail
Fig. 4. The dependence of the packet transfer rate on the total metric
Conclusion
In this paper, a study was made of the dynamic routing protocol, which is an internal routing protocol. The study was carried out taking into account the restrictions introduced in the form of weight coefficients. The default weight values showed their work compared to the addition of another load factor of the package. Based on the results of the study, it was proved that the choice of the route will initially fall on the one that has the smallest mctric and often this way has one transition. Routes with two or more transitions will have a higher metric value, respectively. It is also proved that it is advisable to use the default weighting factors. Change them can be provided that they will be inside one network and not more than one. Also, this approach can be used to optimize adjacent networks and individual data packet flows between them, taking into account the requirements and capabilities of each network.
References
1. IP Routing EIGRP Configuration Guide. (2016). Cisco Systems, Inc. 258 p.
2. Wallace K. (2014). CCNP Routing and Switching ROUTE 300-101 Official Cert Guide. Cisco Press. 1012 p.
3. Snegurov A.V., Chakryan V.Kh. (2015). Improvement of the routing algorithm with load balancing along the paths of unequal cost for the EIGRP protocol. Information processing systems. Kharkov. № 10. Pp. 133-139.
4. Astrakhantsev A.A., Varich V.V., Vakulenko V.S. (2008). Analysis of the coefficients of the EIGRP routing protocol metric. Information processing systems. Kharkov. № 3. Pp. 5-7.
5. Oltfer VG, Olifer N.A. (2006). Computer networks. Principles. Technologies. Protocols. 3rd ed. S.-Pb.: Peter. 958 p. (In Russian)
АНАЛИЗ ЭФФЕКТИВНОСТИ РАБОТЫ ПРОТОКОЛА МАРШРУТИЗАЦИИ БЮКР
Абрамкина Ольга Александровна, Сибирский Государственный Университет Телекоммуникаций и Информатики,
г. Новосибирск, Россия, [email protected]
Аннотация
Рассматривается усовершенствованный дистанционно-векторный протокол выбора маршрута EIGRP в условиях изменения коэффициентов весовой метрики. Весовая метрика протокола EIGRP имеет пять коэффициентов: показатель пропускной способности (KI), загрузки (K2), задержки (K3), показатель надежности (K4) и длина полезной нагрузки (K5). В спецификации Cisco для расчета весовой метрики используются только коэффициенты пропускной способности и задержки, К1 и КЗ соответственно, со значением I, остальные установлены со значением 0. В работе проведено исследование работы сети с изменением значений коэффициентов по умолчанию, так как они могут варьироваться от 0 до 255. Так же проведен анализ сети с включением в весовую метрику коэффициента К2 со значением от 0 до 2.
Доказан тот факт, что при использовании метрики с разными вариациями коэффициентов, теряется часть полезной информации и может произойти ситуация, когда правильно найденный путь не удовлетворяет ограничениям. В результате анализа протокола произведена попытка выбора оптимальных значений коэффициентов для выбранной сети. Для достижения поставленной цели были использованы в качестве сервера - VMware версии 12.5.2 и GNS3 версии 2.0.0b2 в качестве симулятора сети. Исследуемая сеть построена на роутерах модели c7200 с соединениями типа serial.
Результаты моделирования сети с использованием протокола EIGRP в условиях изменения весовых коэффициентов показали, что использование значений коэффициентов по умолчанию целесообразно в том случае, когда нет жестких требований к сети или поиску кратчайшего пути. При возникновении требований к балансировке нагрузки как внутри одной сети, так и между сетями, можно изменять значения коэффициентов и увеличивать их число (включать, например, К2) но не более чем на единицу от значений по умолчанию. Изменения более чем на единицу могут привести к перегрузкам сети при балансировке, а так же невозможности определения кратчайшего пути.
Ключевые слова: протокол EIGRP, коэффициент, маршрут, метрики, ограничения. Литература
1. IP Routing EIGRP Configuration Guide. Cisco Systems, Inc. 2016. 258 p.
2. Wallace K. CCNP Routing and Switching ROUTE 300-101 Official Cert Guide. Cisco Press, 2014. 1012 p.
3. Снегуров А.В., Чакрян В.Х. Усовершенствование алгоритма маршрутизации с балансировкой нагрузки по путям неравнозначной стоимости для протокола EIGRP // Системы обработки информации. Харьков, 2015. № 10. С. 133-139.
4. Астраханцев А.А., Варич В.В., Вакуленко В.С. Анализ коэффициентов метрики протокола маршрутизации EIGRP // Системы обработки информации. Харьков, 2008. № 3. С. 5-7.
5. Олифер В.Г., Олифер Н.А. Компьютерные сети. Принципы. Технологии. Протоколы. 3-е изд. С.-Пб.: Питер, 2006. 958 с.
Информация об авторах:
Абрамкина Ольга Александровна, аспирант, Сибирский Государственный Университет Телекоммуникаций и Информатики, г. Новосибирск, Россия
