Router selection principle
1.1 Basic knowledge of routing
Routing is a relay process that transfers objects from one place to another
The mechanism for learning and maintaining knowledge of network topology is considered a routing function. Transfer data flows through the router and enters the interface
The process of being transferred through the router to the outbound interface is another separate function, considered a switch/forward function. The routing device must have both routing and switching functions to be an effective relay device.
In order to perform routing, the router must know the following three things:
l. The router must determine whether it activates support for the protocol group;
2. The router must know the destination network;
3. The router must know which outgoing interface is the best way to reach the destination.
The routing protocol determines the best path to the destination by metrics. Small metrics represent preferred paths; if two or more paths have the same small metric, then all of them will be shared equally. Splitting data traffic through multiple paths is called load balancing to the destination.
The information required to perform routing operations is contained in the router's routing table, and they are generated by one or more routing protocol processes. The routing table consists of multiple routing entries, each of which indicates the following:
*Learn the mechanism used by this route (dynamic or manual)
*Logical destination
*Management distance
*Measurement value (it is a measure that measures the total "total overhead" of a path)
*The address of the relay device (router) to the destination lower HOP;
*The old and new route information
*Interface associated with the network to go to the destination
Use the command SHOW IP ROUTE to see the above content
The pre-allocation principle of default management distance is: manually set the priority of routing entries higher than dynamically learned routing entries, and the routing protocol with complex routing algorithms with higher priority of routing protocols with complex routing algorithms with simple routing protocols.
Routers generally choose paths with the smallest metric value; if multiple paths with the lowest and same metric value appear at the same time in the IP environment of the CISCO router, load balancing will be enabled on these multiple paths. C ISCO supports 4 paths with the same metric value by default. By using the "maximum-paths" command, it can be recognized that the CISCO router supports up to 6 paths with the same metric value.
RIP is a routing protocol used in small to medium-sized TCP/IP networks. It uses the number of hops as the measure value. Its load balancing function is enabled by default. RIP determines the optimal path without considering bandwidth! ! !
IGRP is a routing protocol used in medium to large TCP/IP networks. It uses composite metrics that take into account bandwidth, latency, reliability, load and maximum transmission unit (M TU), but uses bandwidth and delay values by default. IGRP can also perform load balancing.
After the router starts, it immediately tries to establish a routing relationship with its adjacent routing devices. The purpose of this initial communication is to identify adjacent devices and start communication and learn the network phase structure. The method of establishing neighbor relationships and the initial learning of topology vary with the routing protocol.
The routing protocol exchanges periodic HELLO messages or periodic routing update packets to maintain communication between adjacent devices.
After understanding the network topology structure and the routing table already contains the best path to the known network, forwarding data to these destinations can begin.
The routing protocol that does not send subnet mask information with each network address is called a class selection protocol (RIPv1, IGRP)
When using a class routing protocol, all subnets belonging to the same main class network (Class A, B and C) must use the same subnet mask. A router running a routing protocol with a category routing protocol will perform one of the following tasks to determine the routing network part:
*If the routing update information is about the same main network configured on the receiving interface, the router will use the subnet mask configured on the interface;
*If the routing update is about a network of different main classes that are assigned on the receiving interface, the router will use the default subnet mask according to the address category to which it belongs.
The generation of category inductive routes is automatically processed by category routing protocols.
Classless routing protocols include Open Shortest Path First (OSPF), EIGRP, RIPV2, Intermediate System-to-Intermediate System (IS-IS), and Border Gateway Protocol Version 4 (BGP4).
Using different mask lengths in the same main-class network is called a variable-length subnet mask (VLSM). The classless routing protocol supports VLSM, so the subnet mask can be set more effectively to meet the needs of different subnets for different hosts, and can make full use of the host address.
Most distance vector routing protocols produce regular, routine routing updates that are transmitted only to directly connected routing devices.
In a purely distance vector routing environment, routing updates include a complete routing table. By receiving the full routing table of adjacent devices, the routing can check all known routes and then modify the local routing table based on the received update information. The distance vector method that solves routing problems is sometimes called "routing by rumor"
CISCO IOS supports several distance vector routing protocols, RIPv1, RIPv2 and IGRP. CISCO also directly holds EIGRP, which is an advanced distance vector routing protocol.
The routing protocol is usually associated with the network layer of the protocol group.
Most distance vector routing protocols use Bellman-Ford algorithm to calculate routing. EIGRP is an advanced distance vector routing protocol that uses diffusion correction algorithm (D UAL)
Features RIPv1RIPv2IGRPEIGRP
Count to infinite XXX
Cross-side distance XXXX
Suppress timer XXX
Triggered update, routing reverse XXXX
Load balancing-equal cost path XXXX
Load balancing-non-equal cost path XX
VLSM supports XX
Routing algorithm Bellman-Ford Bellman-Ford Belman-Ford DUAL
Measurement value hop number hop number compound
Jump limit 1515100100
Easy to expand small and medium large
Note: The default hop limit for IGRP and EIGRP is 100, but it can be configured to a maximum of 255.
The link-state routing protocol generates routing update packets only when the network topology changes. When the link state changes, the device that detects the change generates a link state announcement (L SA) about the link (routing). The LSA is then propagated to all neighboring devices through a special multicast address. Each routing device retains an LSA copy and forwards the LSA to its neighbors (this process becomes called diffusion f loading) and then updates its topology database (this is a table containing all link status information for the network). LSA diffusion is used to ensure that all routing devices are aware of this change so that they can update their data and generate an updated routing table that reflects the new network topology.
Features OSPFIS-ISEIGRP
Requires systematic topological structure XX
Keep an understanding of all possible routes XXX
Routing Induction-Artificial XXX
Routing Induction-Automatic X
Event trigger notification XXX
Load balancing-equal cost path XXX
Load balancing-non-equal cost path X
VLSM supports XXX
Routing algorithm DijkstraIS-ISDUAL
Metric link cost (bandwidth) link cost (bandwidth) composite
No hop limit is 1024100
Very easy to expand
The routing processes in each router must leave a single loop-free path to each possible destination logical network. When all routing tables are synchronized and each routing table contains an available route to the destination network, the network reaches a convergence state. Convergence is the activity associated with the routing table synchronization after the network topology changes, such as the state of adding new routes or the existing route changes.
Convergence time is the time required for all routes in the network to achieve consistency in their perception of the current topology structure. The size of the network, the routing protocol used, and numerous configurable timers can all affect the convergence time.
There are two methods of detection:
*When the physical layer or data link layer fails to receive a certain number (usually 3) of continuous keepalive messages, the link is considered invalid.
* When the routing protocol fails to receive a certain number (usually 3) of continuous Hello messages or route updates or similar messages, the link is considered invalid.
Most routing protocols have timers to prevent topological loops from occurring during link state transitions. Article entry: csh Editor in charge: csh
1.1 Basic knowledge of routing
Routing is a relay process that transfers objects from one place to another
The mechanism for learning and maintaining knowledge of network topology is considered a routing function. Transfer data flows through the router and enters the interface
The process of being transferred through the router to the outbound interface is another separate function, considered a switch/forward function. The routing device must have both routing and switching functions to be an effective relay device.
In order to perform routing, the router must know the following three things:
l. The router must determine whether it activates support for the protocol group;
2. The router must know the destination network;
3. The router must know which outgoing interface is the best way to reach the destination.
The routing protocol determines the best path to the destination by metrics. Small metrics represent preferred paths; if two or more paths have the same small metric, then all of them will be shared equally. Splitting data traffic through multiple paths is called load balancing to the destination.
The information required to perform routing operations is contained in the router's routing table, and they are generated by one or more routing protocol processes. The routing table consists of multiple routing entries, each of which indicates the following:
*Learn the mechanism used by this route (dynamic or manual)
*Logical destination
*Management distance
*Measurement value (it is a measure that measures the total "total overhead" of a path)
*The address of the relay device (router) to the destination lower HOP;
*The old and new route information
*Interface associated with the network to go to the destination
Use the command SHOW IP ROUTE to see the above content
The pre-allocation principle of default management distance is: manually set the priority of routing entries higher than dynamically learned routing entries, and the routing protocol with complex routing algorithms with higher priority of routing protocols with complex routing algorithms with simple routing protocols.
Routers generally choose paths with the smallest metric value; if multiple paths with the lowest and same metric value appear at the same time in the IP environment of the CISCO router, load balancing will be enabled on these multiple paths. C ISCO supports 4 paths with the same metric value by default. By using the "maximum-paths" command, it can be recognized that the CISCO router supports up to 6 paths with the same metric value.
RIP is a routing protocol used in small to medium-sized TCP/IP networks. It uses the number of hops as the measure value. Its load balancing function is enabled by default. RIP determines the optimal path without considering bandwidth! ! !
IGRP is a routing protocol used in medium to large TCP/IP networks. It uses composite metrics that take into account bandwidth, latency, reliability, load and maximum transmission unit (M TU), but uses bandwidth and delay values by default. IGRP can also perform load balancing.
After the router starts, it immediately tries to establish a routing relationship with its adjacent routing devices. The purpose of this initial communication is to identify adjacent devices and start communication and learn the network phase structure. The method of establishing neighbor relationships and the initial learning of topology vary with the routing protocol.
The routing protocol exchanges periodic HELLO messages or periodic routing update packets to maintain communication between adjacent devices.
After understanding the network topology structure and the routing table already contains the best path to the known network, forwarding data to these destinations can begin.
1.2 Routing Protocol
Classful routing overview
The routing protocol that does not send subnet mask information with each network address is called a class selection protocol (RIPv1, IGRP)
When using a class routing protocol, all subnets belonging to the same main class network (Class A, B and C) must use the same subnet mask. A router running a routing protocol with a category routing protocol will perform one of the following tasks to determine the routing network part:
*If the routing update information is about the same main network configured on the receiving interface, the router will use the subnet mask configured on the interface;
*If the routing update is about a network of different main classes that are assigned on the receiving interface, the router will use the default subnet mask according to the address category to which it belongs.
The generation of category inductive routes is automatically processed by category routing protocols.
Classless routing overview
Classless routing protocols include Open Shortest Path First (OSPF), EIGRP, RIPV2, Intermediate System-to-Intermediate System (IS-IS), and Border Gateway Protocol Version 4 (BGP4).
Using different mask lengths in the same main-class network is called a variable-length subnet mask (VLSM). The classless routing protocol supports VLSM, so the subnet mask can be set more effectively to meet the needs of different subnets for different hosts, and can make full use of the host address.
Most distance vector routing protocols produce regular, routine routing updates that are transmitted only to directly connected routing devices.
In a purely distance vector routing environment, routing updates include a complete routing table. By receiving the full routing table of adjacent devices, the routing can check all known routes and then modify the local routing table based on the received update information. The distance vector method that solves routing problems is sometimes called "routing by rumor"
CISCO IOS supports several distance vector routing protocols, RIPv1, RIPv2 and IGRP. CISCO also directly holds EIGRP, which is an advanced distance vector routing protocol.
The routing protocol is usually associated with the network layer of the protocol group.
Most distance vector routing protocols use Bellman-Ford algorithm to calculate routing. EIGRP is an advanced distance vector routing protocol that uses diffusion correction algorithm (D UAL)
Comparison of Cisco's IP Distance Vector Routing Protocol
Features RIPv1RIPv2IGRPEIGRP
Count to infinite XXX
Cross-side distance XXXX
Suppress timer XXX
Triggered update, routing reverse XXXX
Load balancing-equal cost path XXXX
Load balancing-non-equal cost path XX
VLSM supports XX
Routing algorithm Bellman-Ford Bellman-Ford Belman-Ford DUAL
Measurement value hop number hop number compound
Jump limit 1515100100
Easy to expand small and medium large
Note: The default hop limit for IGRP and EIGRP is 100, but it can be configured to a maximum of 255.
The link-state routing protocol generates routing update packets only when the network topology changes. When the link state changes, the device that detects the change generates a link state announcement (L SA) about the link (routing). The LSA is then propagated to all neighboring devices through a special multicast address. Each routing device retains an LSA copy and forwards the LSA to its neighbors (this process becomes called diffusion f loading) and then updates its topology database (this is a table containing all link status information for the network). LSA diffusion is used to ensure that all routing devices are aware of this change so that they can update their data and generate an updated routing table that reflects the new network topology.
Comparison of Cisco's Link-state Routing Protocol
Features OSPFIS-ISEIGRP
Requires systematic topological structure XX
Keep an understanding of all possible routes XXX
Routing Induction-Artificial XXX
Routing Induction-Automatic X
Event trigger notification XXX
Load balancing-equal cost path XXX
Load balancing-non-equal cost path X
VLSM supports XXX
Routing algorithm DijkstraIS-ISDUAL
Metric link cost (bandwidth) link cost (bandwidth) composite
No hop limit is 1024100
Very easy to expand
The routing processes in each router must leave a single loop-free path to each possible destination logical network. When all routing tables are synchronized and each routing table contains an available route to the destination network, the network reaches a convergence state. Convergence is the activity associated with the routing table synchronization after the network topology changes, such as the state of adding new routes or the existing route changes.
Convergence time is the time required for all routes in the network to achieve consistency in their perception of the current topology structure. The size of the network, the routing protocol used, and numerous configurable timers can all affect the convergence time.
There are two methods of detection:
*When the physical layer or data link layer fails to receive a certain number (usually 3) of continuous keepalive messages, the link is considered invalid.
* When the routing protocol fails to receive a certain number (usually 3) of continuous Hello messages or route updates or similar messages, the link is considered invalid.
Most routing protocols have timers to prevent topological loops from occurring during link state transitions. Article entry: csh Editor in charge: csh