why subnet was not used in past

[sarahr202]

Hi every body

I   am wondering why  the subnet 0 was not used in past even when classless routing protocol was in use. My book gives the following reason which i don't agree with.

" For example for class b network 150.150.0.0/16, the zero subnet would be 150.150.0.0/24 which creates a bit amiguity at fast glance"

I will try to explain my question with the help of examples below

if we try configure overlapping subnets on router, router will reject the command  with the warning.  But router can accept anoverlapping routes from other neighbors and install it in routing table. For example, a router can have routing table such as:

R 150.150.0.0/16  via 191.191.191.1 s0

C  150.150.0.0/24  directly connected E0

Let say a host,h1 in  150.150.0.0/16  is configured with 150.150.0.1/16 and is located off s0. Let further assume  router receives a packet with dest ip 150.150.0.1 . Router performs  the look up and finds that this  ip address  is within the range of both entries:

R 150.150.0.0/16  s0

C 150.150.0.0/24  e0

Router chose the 150.150.0.0/24 entry because it is the longest match. Of course, h1 will never receives any packet because router will try to send it out e0

The question is would the situation be different if we did not use 150.150.0.0/24? not exactly.  Let me give you another example to make my point.

Let say our router has the entries:

R 150.150.0.0/16 s1

C 150.150.1.0/24 e0

let say host h1 is located off s1 and is configured with 150.150.1.1/24.

Let say router receives a packet destined to 150.150.1.1

Router will try to send the packet out of e0 as a result  host h1 will not receives any packet.  We encounter the same  situation like before when we used zero subnet.  It appears to me  it is not that  zero subnet will cause possible communication , it could be any subnet which overlaps with other entries in routing table could cause the same issue which makes my book claim a moot point.The ambiguity as mentioned in my book could result too if we have overlapping routes in our routing table.  So to say  because of that ambiguity, subnet zero was avoided,  does not seem very convincing.

Any insights would be greatly appreciated.

Thanks and happy new year to all of you.

[Mahesh Gohil · Accepted solution]

Hello Sara,

Yes it will create problem and leads to incorrect routing to wrong path. I was trying in lab and same happens with me

C       150.150.0.0/24 is directly connected, FastEthernet0/1
O       150.150.0.0/16 [110/20] via 10.10.10.2, 00:20:00, FastEthernet0/0

Actually destination is on FE0/0 while router trying to send it on FE0/1.

But this is the problem created by network administrator.  If we design our network properly I don't think we need to face

such routing issue.

well this is my thought and may not be ideal one

Regards

Mahesh

[sarahr202]

Thanks Mashesh

[speculor_cisco]

Hello sarahr202.

I am quite sure that only a scenario with discontigous networks would have your routing tables.

This fact and not zero subnet would cause connectivity problem.

The ambiguity that your book mention is about classful routing protocols when they advertise

a route without network mask: for examble it would not be clear if 172.16.0.0 is the class B

network 172.16.0.0/16 or the subnet 172.16.0.0/24 of the class B network 172.16.0.0/16.

It would be better if someone confirms my first information.

[Peter Paluch]

Hi Sarah,

Regarding the subnet zero and its alleged unusability in classful routing protocols, I've made an attempt some months ago at explaining it here:

https://supportforums.cisco.com/message/3226206#3226206

The entire idea behind my reasoning is that all-zero and all-one subnet are perfectly usable even with classful routing protocols, not to mention classless routing protocols (with classless routing protocols, there are absolutely no limitations on network addresses and netmasks). Avoiding the use of these subnets was driven, in my opinion, mostly by a perception that their processing is inevitably connected with a certain degree of ambiguosity in the classful world (is the network a major network or its first subnet? is the broadcast intended to all stations of the major network or just to stations of the last sbunet?)

Avoiding the use of all-zero and all-one subnets in classless protocols is, in my opinion, technically unfounded and is simply a habit acquired from the classful world, incorrectly applied to classless routing.

The scenario both you and Mahesh described is fine but its nature is not really related to the all-zero subnet but merely to overlapping networks - which is a design illegal both for classful and classless approach to routing.

Best regards,

Peter