Cette page appartient aux archives web de l'EPFL et n'est plus tenue à jour.
This page belongs to EPFL's web archive and is no longer updated.

Exos 2.3

Good Evening!

A few questions regarding ex. 2.3:

1) 2.3.2: Is it Ok for each router to indicate only the /28 subnet(s) which isn't (aren't) on-link as well as the proxy subnet 9.23.30.0/?? for R1 and R2 or do we *have* to indicate a default gateway?

2) 2.3.8: I'm not sure how the bridges will react with the loop. I guessed they will automatically use only one of the possible routes (hopefully the shortest) and therefore the ping packets will only be seen at one observation point each time... is this correct?

Thanks alot,

Christophe Gudin.
Posted by Christophe Gudin on Wednesday 6 December 2006 at 0:01
Comments
1) 2.3.2: Is it Ok for each router to indicate only the /28 subnet(s) which isn't (aren't) on-link as well as the proxy subnet 9.23.30.0/?? for R1 and R2 or do we *have* to indicate a default gateway?

A: each computer should be able to reach any IP address on the figure and in the world. (If this is not enough ask again)

2) 2.3.8: I'm not sure how the bridges will react with the loop. I guessed they will automatically use only one of the possible routes (hopefully the shortest) and therefore the ping packets will only be seen at one observation point each time... is this correct?

A: completely wrong.
Posted by Slavisa Sarafijanovic on Wednesday 6 December 2006 at 11:14
When you do a ping C1 to C2, you've got a table of routes that says you by what path to send the icmp packets.
In this case, you're going to see that C1->B1. B1 is a bridge, by contrast a broadcast occurs, and in sniffer number 1, you can see the icmp packets, with destination mask called M5, next hop M7 and destination prefix 192.168.1.35.

Is it correct?
Posted by Javier Sanchez Tamargo on Wednesday 6 December 2006 at 12:44
1) "the gateway is connected to the Internet only by the means of a web proxy". Does this tell us that any IP trafic passes through the proxy even non-http? In this case is the proxy transparent for non-http traffic? This would force us to use default gateway. But if the proxy only authorizes http trafic than we don't need it. Is my reasoning correct?

2) For Question 8. My reasonig was based on the fact that there is no routing because everything is on the same subnet and routers aren't "real" routers any more because they have only 1 interface. In this case the bridges are the forwarding devices... but I don't know how they forward packets when they had arp answers for C2 for example from 2 interfaces. Could you explain it?

Thanks for your time.

Christophe Gudin.
Posted by Christophe Gudin on Wednesday 6 December 2006 at 13:59
bridges form a tree, so some brigdges will block some of their ports so to have a tree. so one of the observation points will have no traffic, other traffic will go over the tree (tree depends on your choice whice interfaces to block)
Posted by Slavisa Sarafijanovic on Wednesday 6 December 2006 at 16:51
1) "the gateway is connected to the Internet only by the means of a web proxy". Does this tell us that any IP trafic passes through the proxy even non-http? In this case is the proxy transparent for non-http traffic? This would force us to use default gateway. But if the proxy only authorizes http trafic than we don't need it. Is my reasoning correct?

A: you can assume that it is "transparent" or not, as you wish, and your reasoning above is ok. moreover, in each case use of default gw at R1 is the most logical, to save in routing entries, and def gw could be similarly helpful at other routers but is not obligatory at any router in case you assume only http traffic will be used and your browsers are told at which address the proxy is.
Posted by Slavisa Sarafijanovic on Wednesday 6 December 2006 at 16:57
When you do a ping C1 to C2, you've got a table of routes that says you by what path to send the icmp packets.
In this case, you're going to see that C1->B1. B1 is a bridge, by contrast a broadcast occurs, and in sniffer number 1, you can see the icmp packets, with destination mask called M5, next hop M7 and destination prefix 192.168.1.35.

Is it correct?

A: for what figure? M5 is not mask but mac address! (?). In any case, when you ping C2, the destination ("next hop" as you call it, if I understood well) will never be M7. Please us standard names for things...
Posted by Slavisa Sarafijanovic on Wednesday 6 December 2006 at 17:04