Frank H wrote: > > Proper network design? > > I have a few questions for the group that maybe someone can > answer. From my studies when I got CCNA certified, I understood > that different networks were ALWAYS separated by a router. At > my company we have this equipment that was purchased several > months ago that acts as a digital cellular network. It was set > up and was able to operate, but only in a limited way. > Basically, this is the setup - the digital cellular network was > on the 192.168.2.0 subnet (subnet mask 255.255.255.0). The > company development LAN was on the 192.168.0.0 subnet (subnet > mask 255.255.255.0). The two small networks (less than 10 hosts > in each subnet) were all tied together at a 24 port hub. The > gateway to the Internet was through a Linux box. The digital > cellular network was basically a box (with IP address > 192.168.0.100) that passed packets to network 192.168.2.0 > through a low power transmitter to the cellular hosts in the > 192.168.2.0 subnet. With this setup, only one desktop host on > the 192.168.0.0 network could communicate to the 192.168.2.0 > cellular network (desktop host 192.168.0.20). The problem of > only one desktop host in the 192.168.0.0 network being able to > communicate with the 192.168.2.0 network was solved by > replacing the Linux box with a Cisco 2514 router (with two > ethernet interfaces). The configuration for the router was > exactly the same as the Linux box except for one small > addition. The following line was added as a static route: > > ip route 192.168.2.0 255.255.255.0 192.168.0.100 > > Now let me ask you, have you ever seen a router that gets a > packet on one interface pass it right back out the SAME > interface back to another host on that same network?
Sure, it happens all the time. There's nothing non-standard about this. It's quite normal for a router to receive a packet on an interface, look into its routing table, and determine that the packet needs to go back out the same interface it came in on. For example, let's say you have a LAN in Chicago that has two routers on it. One router has a WAN connection to San Francisco and the other router has a WAN connection to New York. Clients on the LAN in Chicago can only be configured with one default gateway. So, let's say that you tell them their default gateway is the router that goes to New York. When the clients send a pcket to San Francisco, the packet goes to the router that connects to New York. That router sends the packet back out the LAN to the router that goes to San Francisco. The router can send an ICMP Redirect to the end host saying essentially "don't use me, use this other router." The host may or may not follow that advice. This is sometimes called "the extra hop problem," although it's not really a problem. In your case, since the cellular box is a bit weird (only supports one host talking through it I think you said), you would probably want to disable ICMP Redirects. >Our setup > basically ties two DIFFERENT class C subnets together through a > hub and the Cisco router makes it all work perfectly. A hub? Now that part is confusing. Are you referring to the cellular box, which sounds like a router to me. It's on two networks, 192.168.0.0 and 192.168.2.0. It's at least a device that can do forwarding based on Layer 3. > This > doesn't sound like standard network design as I've seen it > described in any text so far. I'll describe it a little more > for clarity. If i'm on a desktop PC (IP address 192.168.0.20) > and ping IP address 192.168.2.2, windows will send that packet > to the default gateway (configured as 192.168.0.1 in windows > network applet - which is the Cisco router) since it lies in a > different network (since the subnet mask is 255.255.255.0). The > Cisco router receives this packet destined for the 192.168.2.0 > network and since it matches it with the above static route, > sends it back out the same interface it came in on, back to > another host (192.168.0.100 - the cellular transmitter box) out > to the cellular host (192.168.2.2). That's quite normal. > This is the way the > cellular network equipment manufacturer intended it to work. > The setup works, but it sounds really weird and nonstandard. > Has anyone else encountered such a setup or something similar > before? Is this a kind of network design that is done often? > Doesn't a router normally always route packets from one > interface to another? Depends on the topology. Maybe the router can't get there itself but knows that another router (or host or cellular box or whatever) can. I hope this makes sense. ________________________ Priscilla Oppenheimer http://www.priscilla.com > > Thanks in advance for your responses. > > Frank > Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=49594&t=49536 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
