as always, the answers may be found in the RFC's - specifically in RFC 1812. See in particular pages 93 and 94.
http://www.ietf.org/rfc/rfc1812.txt?number=1812 comments below: ""Priscilla Oppenheimer"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > Wesley wrote: > > > > Heya Priscilla, > > > > Thanks for the explaination. Is it safe to assume that Cisco > > routers do not > > perform an all subnets broadcast? to quote the RFC: "In a CIDR routing domain, wherein classical IP network numbers are meaningless, the concept of an all-subnets-directed-broadcast is also meaningless. To the knowledge of the working group, the facility was never implemented or deployed, and is now relegated to the dustbin of history." in other words - some pretty smart folks have already been there, done that, and concluded it weren't worth the effort > > I've never seen them do it, but I can't say for sure that they never do. to quote the RFC: "The first version of this memo described an algorithm for distributing a directed broadcast to all the subnets of a classical network number. This algorithm was stated to be "broken," and certain failure cases were specified." the reference seems to be to RFC 1009, which does indeed discuss directed broadcasts and certain exceptions. > > > I found something off Google > > that I would > > like to share with you guys. This is an excerpt from "TCP/IP > > Tutorial" by > > IBM. no doubt written by folks who studied the RFC's > > Good stuff. Thanks. > > > > > All-Subnets-Directed Broadcast Address > > > > If the network number is a valid network number, the network is > > subnetted > > and the local part is all ones (for example, 128.2.255.255), > > then the > > address refers to all hosts on all subnets in the specified > > network. In > > principle routers may propagate broadcasts for all subnets but > > are not > > required to do so. In practice, they do not; there are few > > circumstances > > where such a broadcast would be desirable, and it can lead to > > problems, > > particularly if a host has been incorrectly configured with no > > subnet mask. > > Consider the wasted resource involved if a host 9.180.214.114 > > in the > > subnetted Class A network 9 thought that it was not subnetted > > and used > > 9.255.255.255 as a local broadcast address instead of > > 9.180.214.255 and all > > of the routers in the network respected the request to forward > > the request > > to all clients. If routers do respect all-subnets-directed > > broadcast > > address, they use an algorithm called reverse path forwarding > > to prevent > > the broadcast messages from multiplying out of control. See RFC > > 922 for more > > details on this algorithm. again, if I read my RFC history correctly, this avenue proved more difficult than at first believed, and was eventually abandoned. > > > > I guess an important point here is whether routers respect the > > all-subnets > > directed broadcast. no longer important. it isn't done. > > > > A question on the /32 entries in the routing table. I can > > understand that > > each async line can only have one host at the remote end and > > therefore would > > have a host route in the routing table. So how do the hosts on > > async lines > > receive broadcasts? > > Why would the hosts need to receive broadcasts? A broadcast means go to > everyone. Well, "everyone" is just one device in this case! ;-) There's no > need for anyone other than possibly the router to find the MAC address of > these devices, so there's no need for ARP broadcasts (which may not have any > meaning for async anyway). I suspect that the point here is that in an asynch world there might still be applications that do broadcast, and that all hosts in the same "subnet" e.g. 192.168.1.x receive that broadcast, even if every host is asynch, and every host was in theory a /32 let's say, for example, I'm a brokerage firm, and a lot of my brokers work from home. they dial in, and they run their quote application. the quote server sends nothing but broadcasts. > > You wouldn't want a host on an async link to be announcing things like the > existance of its local printer, so there's little need for service > announcement-type broadcasts. If you did need service announcements to get > forwarded, you could use a helper address. > > DHCP broadcasts probably aren't necessary. (A lot of async lines use PPP. > The IP PPP NCP does IP address assignment). If you did need DHCP, then you > could use a helper address. > > > What would be their broadcast address? > > I guess it would be the same as the actual address if there were a need for > it, but there may not be any need for it. > > > It > > would only > > make sense that the broadcast packet was not delivered to the > > hosts > > connected to Router 5. That is why the packets kept bouncing > > back and forth. > > Otherwise it would have been delivered to the hosts and this > > problem would > > not have surfaced. > > > > Why is it that the all ones subnet was initially excluded as a > > valid subnet? > > Paranoia?? ;-) Prisoner: Who are you? Voice: I am the new Number 2 Prisoner: Who is Number 1? Voice: You are, Number 6. My guess is that is has more to do with the old original routing algorythms. The answer might be found in the RFC's, or they might be found in the archives of long since defunct router makers and router code writers. it is certainly far enough past my bedtime that I don't want to start to research. > > And that's all I have to say on the subject. > > Priscilla > > > > It would seem perfectly OK to me to use the all zeros and all > > ones subnet > > numbers. I mean with prefix routing, these things can be > > distinguished. Like > > you said, the problem discussed in the article was pretty > > 'artsy'. > > > > As always, thank you so very much. > > > > Wesley > > > > ""Priscilla Oppenheimer"" wrote in > > message > > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > > Wesley wrote: > > > > > > > > So there isn't a broadcast address for all /27 subnets? > > > > > > I don't think sending to all subnets of a network is > > something that IP > > ever > > > defined. > > > > > > > I > > > > basically > > > > understand that the last address of each subnet is reserved > > for > > > > subnet > > > > broadcast. I was just wondering if the broadcasting > > > > architecture allowed for > > > > all subnets to be broadcasted at once. And Mark, since you > > are > > > > the only one > > > > replying mind if you check out the CCO link in the original > > > > > > I hope Mark will answer too, but since we're the only ones > > talking now, > > I'll > > > jump in. ;-) > > > > > > > post and tell me > > > > your views on the issues that I have highlighted. I'll > > provide > > > > the link > > > > again > > > > > > > > http://www.cisco.com/warp/public/105/40.html > > > > > > I think the main thing to realize about the article is that > > it's a very > > > strange case. Notice that the Asynch routers have a bunch of > > host-specific > > > routes (/32). And then their E0's are configured with a /24 > > subnet mask, > > > even though they probably should really be /26 to fit the > > network design. > > > > > > And then to make the problem happen they had to have a host > > misconfigured > > > for /24 also and have it send a NetBIOS (or other) broadcast > > to x.x.x.255. > > > > > > I suggest that you set up a more normal situation in your lab > > and see if > > you > > > can get the problem to happen. Perhaps TAC ran into a problem > > matching the > > > scenario they describe. But is the problem reproducible under > > more normal > > > condistions? (Perhaps TAC just made up the scenario too!? > > There are parts > > of > > > it that aren't too believable. ;-) > > > > > > Please see a few more comments below. > > > > > > snip > > > > > > > > > > > > > > > I was going thru this article about the effect of using > > the > > > > all ones > > > > > > subnet. > > > > > > There are somethings that I'm still confused about. The > > > > link is > > > > > > > > > > > > http://www.cisco.com/warp/public/105/40.html > > > > > > > > > > > > 1. In the first example, when host 195.1.1.24 sends a > > local > > > > broadcast > > > > > to > > > > > > 195.1.1.255, will hosts attached to router 2's async > > lines > > > > receive the > > > > > > broadcast? > > > > > > No, the asynch lines are using /32. > > > > > > > > > > > > > > > 2. OK, its a directed broadcast and router 2 looks up > > its > > > > > > I don't think Router 2 thinks it's a directed broadcast. The > > destination > > > address doesn't match any of the /32 host routes, so Router 2 > > sends the > > > packet out the default route. > > > > > > > routing > > > > > table > > > > > > and > > > > > > forwards it out using the default route. Router 1 > > receives > > > > the packet. > > > > > I > > > > > > believe the packet is forwarded out to all 192.1.1.x/26 > > > > subnets, > > > > > right? > > > > > > No, not all subnets. Router 1 has a specific route for subnet > > 192.1.1.192. > > > (11000000 in the last octet). If a packet comes into that > > subnet, it's > > > supposed to go to Router 5. See the static route that points > > to Router 5 > > > (195.1.2.5). > > > > > > Now, Router 1 should recognize that the incoming packet is a > > directed > > > broadcast for subnet 192 and not forward it if "no ip > > directed-broadcast" > > is > > > configured, which is the default these days. > > > > > > > > > Will > > > > > > Router 1 forward the packet back to Router 2? I hope > > not > > > > > > > > > > > > No. > > > > > > > > > 2a. Another way of looking at it is router 1 thinks > > that it > > > > is a > > > > > > broadcast > > > > > > only for subnet 195.1.1.192 and forwards it out only to > > > > router 5. > > > > > > That's my interpretation too. > > > > > > > > Hmmm > > > > > > .... > > > > > > I'm definitely confused > > > > > > > > > > > > 3. Router 5 receives the packet from router 1. How will > > it > > > > interpret > > > > > the > > > > > > packet? I'm guessing that the router sees it as a > > directed > > > > broadcast > > > > > > Router 5, like Router 2, has a bunch of /32 host routes. The > > incoming > > packet > > > doesn't match any of those, so Router 5 sends it out the > > default route. > > > > > > > > and > > > > > > send it out via the default route. Is it normal that > > > > routers forward a > > > > > > packet out from an interface that it received on? > > > > > > Well, not too common, but it does happen sometimes. > > > > > > > As in its > > > > received > > > > > on > > > > > > e0 > > > > > > and forwarded out e0 as well > > > > > > > > > > > > 4. Once router 1 receives the packet from router 5, > > will it > > > > forward > > > > > the > > > > > > packet out to all 192.1.1.x/26 subnets again or just to > > > > router 5. > > > > > > Just to Router 5 > > > > > > > The > > > > > > article did not detail this part and just specified > > that it > > > > will > > > > > bounce > > > > > > between routers 1 and 5. It also says that routers 2 > > thru 4 > > > > see the > > > > > > 'broadcast' only once. The way I see it , if all subnets > > > > receive the > > > > > > broadcast then routers 2 thru 4 should receive the > > packets > > > > as many > > > > > times > > > > > > as > > > > > > router 5. > > > > > > > > > > > > I would appreciate all the help I can get. I know you > > gurus > > > > can help > > > > > me > > > > > > out. > > > > > > Thanks!! > > > > > > > > > > > > Wes Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=49103&t=48996 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]