gopstein@sisun1.squibb.com (Rich Gopstein) (11/26/90)
We currently have a flat (non-subnetted) class-B network connected via bridges. At the moment, the number of machines talking IP is less than 150, so they are all numbered 140.176.0.x. We would like to start subnetting our network with a subnet mask of 255.255.255.0 by replacing one of the bridges with a router, changing the netmask, and renumbering the machines on the new subnet to 140.176.1.x (leaving the machines on the rest of the net numbered 140.176.0.x). The problem is that our cisco technical contact (in the local sales office) claims that we have to renumber all of the 140.176.0.x machines since zero is not legal in the subnetted portion of the address (ie the third octet)... We asked him whether he meant the HOST portion of the address, but he said that he really meant the third octet. This doesn't make sense to me. Is there a reason that a zero in the subnetted portion of the address will cause cisco routers a problem? Rich Gopstein gopstein@squibb.com rutgers!squibb.com!gopstein -- Rich Gopstein gopstein@squibb.com .!rutgers!gopstein%squibb.com
JOHN@heap.cisco.com (John Wright) (11/27/90)
To: gopstein@sisun1.squibb.com.#Internet In-Reply-To: <1120@soleil.UUCP> You can use subnet 0, but we advise against doing so, in particular when you have hosts which are using the old style 0.0.0.0 broadcast addresses. service subnet-zero will enable subnet zero on a cisco. John Wright Customer Engineering cisco Systems, Inc.
jfp@INEL.GOV (Jeff Pack) (11/27/90)
In article <1120@soleil.UUCP>, gopstein@sisun1.squibb.com (Rich Gopstein) writes: |> zero is not legal in the subnetted portion of the address (ie the third |> octet)... |> |> This doesn't make sense to me. Is there a reason that a zero in the |> subnetted portion of the address will cause cisco routers a problem? |> |> You should be able to make the cisco recognize the zero subnet by adding the following to the configuration: service subnet zero As far as the legalities are concerned, the RFC's are a permissive standard. For example, most other vendors agree that broadcast is all ones in the octet (i.e.; 255) but Sun configures their default broadcast to be all zeros. They say it depends how you interpret the RFC. I think something similar is happening with subnet zero.... Jeff --------------------------------------------------------------- Jeff Pack UUCP: ...!uunet!inel.gov!jfp Idaho National Engineering Lab Internet: jfp@inel.gov P. O. Box 1625 M.S. 2603 Phone: (208) 526-0007 Idaho Falls, ID 83415 FAX: (208) 526-9936 ========== long legal disclaimer follows, press n to skip =========== ^L Neither the United States Government or the Idaho National Engineering Laboratory or any of their employees, makes any warranty, whatsoever, implied, or assumes any legal liability or responsibility regarding any information, disclosed, or represents that its use would not infringe privately owned rights. No specific reference constitutes or implies endorsement, recommendation, or favoring by the United States Government or the Idaho National Engineering Laboratory. The views and opinions expressed herein do not necessarily reflect those of the United States Government or the Idaho National Engineering Laboratory, and shall not be used for advertising or product endorsement purposes.
albright@dirt.cisco.com (Bob Albrightson) (11/27/90)
> This doesn't make sense to me. Is there a reason that a zero in the > subnetted portion of the address will cause cisco routers a problem? Use the config command (before any interface commands): service subnet-zero -bob
fortinp@bwdls56.bnr.ca (Pierre Fortin) (11/27/90)
In article <1120@soleil.UUCP>, gopstein@sisun1.squibb.com (Rich Gopstein) writes: > We currently have a flat (non-subnetted) class-B network connected > via bridges. At the moment, the number of machines talking IP is > less than 150, so they are all numbered 140.176.0.x. We would like > to start subnetting our network with a subnet mask of 255.255.255.0 > by replacing one of the bridges with a router, changing the netmask, > and renumbering the machines on the new subnet to 140.176.1.x (leaving > the machines on the rest of the net numbered 140.176.0.x). > Ooooohhhh.... I guess small *really is* beautiful; read on... > The problem is that our cisco technical contact (in the local sales office) > claims that we have to renumber all of the 140.176.0.x machines since > zero is not legal in the subnetted portion of the address (ie the third > octet)... We asked him whether he meant the HOST portion of the address, > but he said that he really meant the third octet. That's correct, subnet zero is no longer legal, but it used to be. I would suggest that your local rep RTFM... All you have to do is code service subnet-zero before you attempt to change the subnet mask. Doing it the other way around will not be accepted by the cisco. However, you *should* consider renumbering those nodes on subnet zero; don't know if cisco will support this feature(?) (backward compatibility) for much longer. Consider yourself lucky you only had a Class B network... We are almost finished our Class A conversion (you guessed it! Subnet *zero* AND about 2000 nodes when we started. The worst part: while converting old nodes, there were new nodes being added to this old network. Anyway, 1991 will free us of this costly startup mistake. > > This doesn't make sense to me. Is there a reason that a zero in the > subnetted portion of the address will cause cisco routers a problem? > No problem with the zero, but like I said above, seriously consider an orderly conversion. > > Rich Gopstein > gopstein@squibb.com > rutgers!squibb.com!gopstein > Good luck with your cisco router(s), Pierre Fortin Bell-Northern Research I know, my postings are Internet Systems P.O.Box 3511, Stn C terse and humourless. So? (613)763-2598 Ottawa, Ontario RIP: aptly named protocol fortinp@bnr.ca Canada K1Y 4H7 AppleTalk: Adam&Eve's design
edb@crg8.sequent.com (Edward Bunch) (11/30/90)
I've heard this subnet 0 stuff before but I don't understand *exactly* what they mean by it. I can see the example of a class B being subnetted 8 bits pretty clearly. ( 138.95.X.Y ; X nor Y can be 0 ) Y I really understand, no hosts can be 0. X isn't so clear. Doesn't all this 138.95 stuff on the left keep the network from being 0? What if we are doing something weird like subnetting 9 bits? Which address configurations do I avoid now? Ed
JOHN@heap.cisco.com (John Wright) (11/30/90)
Ed,
To use your example of class B 138.95.0.0 with subnet mask 255.255.255.128
or 138.95.0.0 with 9 bits of subnetting.
A subnet of zero is any host address with the subnetting bits set to
zero, or rather hosts 138.95.0.1 thru 138.95.0.126. All of these
addresses are on 'subnet zero'. The address 138.95.0.0 is not a host
address for obvious reasons, the address 138.95.0.127 is not because
with a subnet mask of 9 bits this is a broadcast address to subnet zero,
when using the all ones form of the broadcast.
Let's concentrate most significant two octets of the internet address
(as 138.95 won't change) our subnet mask of 255.128 is:
11111111 10000000
138.95.0.1 would have right two octets 00000000 00000001
the first normal subnet in this case would be 00000000 10000000 the first
real host then would be 00000000 10000001 or 138.95.0.129. The address
138.95.0.255 with nine bits subnetting becomes a directed broadcast at the
128 subnet or 00000000 1 1111111
^ ^^^^^^^
subnet host
See the old gateway manual chapter 11, the 8.1 gateway manual chapter 5,
Douglas Comer's book on TCP/IP, or RFC 950 for more details.
John Wright
Customer Engineering
cisco Systems, Inc.fortinp@bwdls56.bnr.ca (Pierre Fortin) (12/01/90)
In article <20599@crg5.UUCP>, edb@crg8.sequent.com (Edward Bunch) writes: > > I've heard this subnet 0 stuff before but I don't understand *exactly* > what they mean by it. I can see the example of a class B being > subnetted 8 bits pretty clearly. > > ( 138.95.X.Y ; X nor Y can be 0 ) > > Y I really understand, no hosts can be 0. X isn't so clear. Doesn't > all this 138.95 stuff on the left keep the network from being 0? The issue is *SUB*net 0, therefore, in your example, that's X and X alone. > > What if we are doing something weird like subnetting 9 bits? > Which address configurations do I avoid now? It's all quite simple if you convert everything to bits. With a 9/7 split, 138.95.0.0 is still obvious, but 138.95.0.128 may not be. This one is illegal because the Y (7 bits) is 0. Its X is actually 00000000.1xxxxxxx; while 138.95.0.129 is the first legal address: 00000000.1xxxxxxx for subnet and xxxxxxxx.x0000001 for host. If you follow this through, you will find that the legal ranges are: 138.95.0.129 - 138.95.0.254 (138.95.0.255 is subnet 1's broadcast) 138.95.1.1 - 138.95.1.126 138.95.1.129 - 138.95.1.254 138.95.2.1 - 138.95.2.126 138.95.2.129 - 138.95.2.254 . . . 138.95.254.1 - 138.95.254.126 138.95.254.127 - 138.95.254.254 138.95.255.1 - 138.95.255.126 THE END... Addresses beyond would result in the subnet being all 1's. My rule is very simple: always convert to binary and look for all 1's or all 0's in either field. The RFCs actually allow for non-contiguous bits to identify the subnet. Consider the case where the subnet mask for 138.95.x.x is 11111111.11111111.11xx11x1.xxx1xx11; if you can work this one out, you will understand fully. Well, actually consider why either the subnet portion AND the host portion of the address can NEVER be less that 2 bits each if subnetting is in use. Now you understand it all... :^) > > Ed Happy bits, Pierre Fortin Bell-Northern Research I know, my postings are Internet Systems P.O.Box 3511, Stn C terse and humourless. So? (613)763-2598 Ottawa, Ontario RIP: aptly named protocol fortinp@bnr.ca Canada K1Y 4H7 AppleTalk: Adam&Eve's design