I was lazy and found various reasons for myself to "drag the draft". Today, I am finally unable to believe it. At least I have to make up my previous article today.
As mentioned before, to connect different CIDR blocks in different regions, you need to use routers or layer-3 switches to use their routing functions to let data packets know how to reach their destination between different IP address segments. However, each vro only knows the IP network segment directly connected to itself, but does not know its next door and its next door. The method previously mentioned is to manually specify the IP address segment to which a packet is to be dropped.
However, it is too troublesome to manually set static routes on each vro. As the number of vrouters increases, the number of inputs increases exponentially. Second, the entire network may not be able to run even if one or two numbers are wrong during configuration. In addition, troubleshooting may cause exceptional troubles. Static Routing is a solution to connectivity. But it is a thing from nothing. Configuration is troublesome, error-prone, and difficult to manage. As a result, some people may think that it would be nice to have a protocol to automatically let the routers learn the routes that are directly connected to each other. The predecessors are really powerful. They have also developed such protocols to allow routes to learn from each other before, and dynamically adjust the topology according to the current status of the router. An Optimal Path is selected based on the actual situation. These protocols are called dynamic routing protocols.
There are many kinds of dynamic routing protocols, and there are many ways to divide them according to different classification methods. But the most important way is to divide them into two types: "distance vector" and "link state. From the perspective of information exchange, Distance Vector means that the router running the Distance Vector Routing Protocol shares all the routing information it knows with its neighbors, but only shares the information with the directly connected neighbors. The link status is like this. The router that runs the link status routing protocol only shares the status of the directly connected link with the neighbor. This neighbor refers to a domain ), or all routers in an area. There is a metaphor in Baidu, which may not be accurate, but is very vivid. "One is the shortest distance, and the other is the shortest time. It is like distance vector and link status. The distance is short and useful. It's all about car congestion. You haven't had to block it for a few hours. It's the most practical to drive a wide car on the road. (Http://zhidao.baidu.com/question/403285014.html )".
Although there are many dynamic routing protocols, their basic working principle is still not good. Exchange protocol data packets to exchange route information that is known to each other. Then, based on some algorithms, we can use the content in these protocol packets to calculate an optimal path. Then, it is thrown into the "route table" for the router to call. In some cases, if the route information changes, the "route table" will be updated ". When you fail to receive the protocol data packet of a route after a certain period of time, you will think that the router is not working, and then remove the router from the protocol, re-calculate the route of the entire network.
This is a magical process.
The specific routing protocol will be detailed later.
650) this. width = 650; "originheight =" 854 "originwidth =" 480 "src =" http://a4.qpic.cn/psb? /D43ad4cf-8d96-4299-a205-359c1c758a27/CABEHrT6dO3d0arXGywbt7OpZMalzK6YKYpazrGn31Q! /B/dBgtDZj0AwAA & bo = 4 AFWAwAAAAABAJM! "Idx =" 0 "style =" margin: 0px; padding: 0px; border-width: 0px; border-style: none; vertical-align: top; width: 300px; height: 534px; "/>
When will the next seven days of rain in Guangzhou end...
