subnetting question

dsealey · ‎08-20-2002

I've got a client that changed the subnet mask of his class C network to 255.255.252.0. It was 255.255.255.0. The address range that we were using was 192.168.253.1 - 192.168.253.254. Now the client is trying to use the 192.168.252.1 - 192.168.255.254 range. This just does'nt look right to me.

Has anyone got any thoughts on it ?

Prashanth Krishnappa · ‎08-21-2002

Use the subnet calculator

http://www.cisco.com/cgi-bin/Support/IpSubnet/home.pl

dsealey · ‎08-21-2002

The subnet calculator was the first thing I went for when I saw this. It returns an invalid mask for 192.168.252.0 255.255.252.0 . Everyone I talk to seems to think it should work though. The only reason I have a problem with it at all is that the equipment I have on the customers network does'nt work correctly with this new mask. :(

tim.hunt · ‎08-22-2002

When you say not correctly what exactly is it doing? Remember that everything on the segment has to have the same Subnet mask. If one machine thinks it is on a bigger subnet than it actually is it will not correctly send packets to the gateway when it should.

TH

jfraasch · ‎10-15-2002

If you change your address range to a class B it will work. Use 172.16.252.0/22. This will give you the range that you are looking for: 172.16.252.1-172.16.255.254.

And like the other reply said, be sure all the devices on the 'new' network have the SAME mask.

That will definitely work, maybe a bit more work than what you had in mind, but it will work.

James

jfraasch · ‎10-15-2002

By the way, the reason I am assuming that it didnt work for you is because you may be doing Classful routing. If you have older equipment then it may not support Class-less addressing. It needs to or else it will recognize your 192.168.252.x as a class C ONLY. With classless routing it will recognize it as whatever you want it to.

I hope that makes sense. So either your equipment needs to be updated to support classless addresses or you need to change your Address scheme to a class B.

James

nrgizer2 · ‎10-16-2002

Hi...

As a newcomer to the world of Cisco internetworking I am slightly confused by this subnet mask question.

I thought the whole point of subnetting was to create many networks instead of just the one that the default subnet mask of Class A,B,C IP address provides?

You can't take bits away from the default 255.255.255.0 of Class C networks can you? Only add them?

scottmac · ‎10-16-2002

Yes, you can "subtract" subnet mask bits in certain scenarios. The device must be able to understand classless masks, and the routing protocol must also support classless addresses/masks.

The purpose of doing so is to "supernet:" which is basically summarizing a group of routes down to one address. Supernetting reduces the size of the routing/forwarding tables for more efficient resource utilization. Advertising one route (which represents a group of many more) is also bandwidth efficient for WAN links.

You may find more information by searching around for "Classless InterDomain Routing" (CIDR, or CIDR blocks).

FWIW

Scott

jamesleebell · ‎10-31-2002

Sure you can. It's called supernetting. A previous employer used it to have one subnet for one heavily populated floor. They had a range of ~30 class C addresses with 3rd octets ranging from 64 to 95. They took the two class C subnets 66.0 and 67.0 and supernetted them by changing the mask from 255.255.255.0 to 255.255.254.0. This gave one subnet.

Subnet = 66.0 = (0100001/0.00000000)

First ip = 66.1

Broadcast = (0100001/1.11111111) = 67.255

Last ip = 67.254

Note that this also - as long as you had classless ip routing turned on - meant that 67.0 was a valid host address.

jamesleebell · ‎10-31-2002

As to your original question, the ranges are correct so far as classless routing is rigged. Looking at the binary calcs for 192.168.253.0 supernetted back to 22 bits...

253.0 = (111111/01.00000000)

so 253 can't be a network, since by definition the network has host bits of all zeros. Zero this out and you get the network and other info:

network = (111111/00.00000000) = 252.0

first ip = 252.1

broadcast = (111111/11.11111111) = 255.255

last ip = 255.254

range = 192.168.252.1 -> 192.168.255.254

Which is what you referenced in your post. Just for the heck of it, working backwards for the subnet just before this:

last subnets broadcast = 251.255 (i.e. the one just before the above network)

last valid ip = 251.254 (one digit before broadcast)

251 = 111110/11 (with the slash showing the effect of the supernet mask)

by def. previous network has hosts of all zeros, so...

network = 111110/00.00000000 = 248.0

first ip = 248.1

range = 192.168.248.1 -> 192.168.251.254

So there's the last two subnets of a class C supernetted by two bits.

e.muyalde · ‎01-22-2003

Hi,

You are right about that, but in this situation he is not subnetting, instead he is summrizing four class C network into one. Which is called Supernetting.

For example, 192.168.1.0/24 to 192.168.254.0/24 is 254 class C network s that can be summurized and advertised as one network by changing the default subnet mask from /24 to /16. ( 192.168.0.0 255.255.0.0)

b.go · ‎10-24-2002

I've 4 buildings with address 10.252.8.0 - 10.252.15.0 mask=255.255.248.0 . I would like to divide them to one network per building. 10.252.8.0 for server area, 10.252.9.0 for 1st building and etc by changing the mask to 255.255.255.0. I assign one port for each vlan per building (e.g: 10.252.9.1 mask=255.255.255.0), and configure the client gateway pointing to this interface.

1. Is my client able to access the server area (10.252.8.x mask 255.255.255.0) with ip e.g: 10.252.9.10 mask=255.255.248.0 (old one, which I thought it shoud not be able to access since it is in the different network). ?

2. If my client can access servers by using old mask (21 bits), and changing the ip into another 10.252.10.10 mask still using 21 bits. Would it be able to access the server area ?

Thank you for your advices.

Best regards

BUDI

mklaphek · ‎11-04-2002

I'm not sure that I understand you correctly, but it sounds like it's working because the router supports Proxy ARP. In other words, it knows the subnets connected to it and can respond to an ARP of a client with a misconfigured subnet mask.

b.go · ‎11-08-2002

I'm sorry for not giving enough information, and thank you for your response.

Previously it was a flat network (10.252.8.0 - 10.252.15.0 with mask 255.255.248.0) with around 400 users. I tried to divide them without changing clients ip addresses. Luckily the clients ip addresses had been arranged like:

1st buildling 10.252.9.x mask 255.255.248.0, 2nd building 10.252.10.x mask 255.255.248.0, etc. So I changed only their mask to 255.255.255.0.

The interface(on L3 switch) for each network are: 10.252.9.1/24 , 10.252.10.1/24, etc.

1. I conducted a testing by not changing a client netmask (still 21 bits), and it is still able to access the proxy/router on the other network (10.252.8.x/24).

I wonder whether the changes that I have made will reduce the broadcast ? (since they are still able to access the other part of network) ?

2. I am thinking that the addressing is a little bit confusing for me. Since the new mask (24bits)is on 'lower' level compared to (21bits), does it cause the clients with old mask are still able to access 24bits network ?

Thank you very much.

Best regards,

Budi