English Deutsch Français Italiano Español Português 繁體中文 Bahasa Indonesia Tiếng Việt ภาษาไทย
All categories

2 answers

Actually Poison Reverse is used with router protocols to avoid looping packets around several routers.

All routers have routing tables that they keep up to date by talking to other routers. They broadcast these tables so that other routers know how to get to networks the routers are connected to.

Say we have three routers all hooked to each other, and router A has a link to 10.1.1.1. Routers B and C see that router A has a link to 10.1.1.1 and update their tables so that if they have a packet destined to 10.1.1.1, they will send it to router A. But they also know to send it directly to router A because they keep track of how many routers(also knows as hops or a metric) it will take to get to 10.1.1.1. So router A has a metric of 1, and routers B and C have a metric of 2.

Now router A loses the link to 10.1.1.1. It needs a new route, but if it asks routers B and C, they say they have a metric of 2. Router A doesn't know that the metric is through router A, so it starts sending packets to B and C, which then send the packets back to A. This starts a loop.

So instead, when it loses the link to 10.1.1.1, router A will send a routing table with a metric of 16 to all routers, which is considered an infinite link and routers B and C won't send packets for 10.1.1.1 to router A. This is called route poisoning.

Now what happens if routers A and C transmit their routing tables at the same time? Router A thinks its got a good route to 10.1.1.1, while router C doesn't think router A is a good link. So on the next update, router C will think that router A has a good link after the next update, starting the loop mentioned earlier. But with reverse poisoning, when router C gets the message that router A isn't a good link anymore, it will immediately transmit that information to A, so that A doesn't think that router C has a good link.

There are probably better places online with diagrams on how all of this works instead of this little quickie :)

2006-08-04 15:25:31 · answer #1 · answered by Bryan A 5 · 0 0

Ok, listen, I couldn't answer your question offhand, so I started doing some research. I'll just give you the link, but it looks like poison reverse is sent to the upstream router and dependencies on that router are marked which results in a truncated broadcast tree which avoids duplicates.
ur
Course, I'm probably wrong...lol

http://64.233.161.104/search?q=cache:6OQ5YsSGVX0J:www.ncne.nlanr.net/documentation/faq/mcast_eng_faq.html+How+Poison+Reverse+multicast+packet+can+be+used+to+avoid+duplicated+IP+multicast%3F&hl=en&gl=us&ct=clnk&cd=1&client=firefox-a
http://64.233.161.104/search?q=cache:8-CGZnudXbIJ:www.cisco.com/networkers/nw00/pres/2214.pdf+How+Poison+Reverse+multicast+packet+can+be+used+to+avoid+duplicated+IP+multicast%3F&hl=en&gl=us&ct=clnk&cd=5&client=firefox-a

Enjoy. Because that made my eyes blleed and normally I like reading technical stuff.

2006-08-04 15:15:45 · answer #2 · answered by Christine G. 2 · 0 0

fedest.com, questions and answers