Subnetting.  A simple thing that fills many with dread.  The swirl of numbers flying about when discussing subnetted networks can make your head respond in kind.  If subnetting wasn’t challenging enough we have long dealt with this thing, this ’subnet-zero’ thing.  Discussion on the topic (this post included) lingers for some unknown reason.  It vexes me.  I even read a recent post (written about a month ago) that suggested NOT using it was still a viable concept.  At the risk of taunting the author I shall refrain from links to that post.  I can’t tell you how much I disagree with such a statement.  Well, scratch that.  I actually can tell you how much I disagree.  Please enjoy:

First, a review:  What is ‘ip subnet-zero’?

It’s a command you enter into a Cisco router (or L3 switch), actually.  From global configuration mode you have two choices:  ip subnet-zero or no ip subnet-zero.  If you want to use this thing we call subnet zero you enter the former.  If you don’t want to use it, enter the latter.  And I don’t understand you if you don’t want to use it.  Cisco doesn’t either.  They have been telling you for a long time that you should be using it.  The command has been enabled by default since IOS 12.0 which has been out for the better part of a decade.  To stop using it you would have to intentionally go into the router and disable it (e.g. no ip subnet-zero).  And who are you to disable a Cisco default?  Hmmph.  I thought so.

So it’s a command.  Cool.  But what does it actually do?  In simple terms the command controls whether or not the all zeroes and all ones subnets are valid.  Easy enough to say but it’s a little more involved to understand.  Let me explain by way of example.  Here is an example of a simple subnetting problem that uses subnet zero (and yes, I’m assuming you already have a little bit of subnetting skill):

In this example we take the 192.168.44.0/24 network and subnet it by 4 bits,  dividing it into 16 networks.  If we look at the newly created networks a little more closely we will see something interesting.  Using the image below, notice that the first network has all of the subnet bits set to zero.  Also notice that the last network has all of the subnets bits set to one.  We call the network with the subnet bits set to zero the “all zeroes subnet”.  We call the network with the subnet bits set to one the “all one’s subnet”.  Pretty clever names, I know.  Collectively we refer to both of them as subnet-zero.  Yeah, that makes sense?  Deal with the weirdness.

Two things you need to notice:

First:  To the user, the all zeroes subnet looks alamringly like the original network.  The original network was 192.168.44.0/24.  The all zeroes subnet is 192.168.44.0/28.  Does that creep you out?

Second:  The all one’s subnet has a decimal value equal to the subnet mask value in the last octet AND the broadcast address of the all zero’s subnet is 192.168.44.255 which is identical to the broadcast address of the original network (192.168.44.0/24).   How about that?  Feeling spiders crawling up your legs?

What’s my point?

First:  People got confused by the striking similarity between 192.168.44.0/24 and 192.168.44.0/28.  Let me restate that in case you didn’t catch it.  People got confused by… (never mind the rest, it doesn’t matter).  The important word, in case you missed the italics, is PEOPLE.  Computers, routers and other network devices were NEVER confused by the apparent similarity.  To the computer, a device which thinks only in binary, 192.168.44.0/24 is just as similar to 192.168.44.0/28 as 11.12.0.0/14 is.  The difference is plain to see for the computer.  The same can’t be said for the fragile mental stability of the network administrator.  Because people were confused the decision was made long ago to simply throw out the all zeroes subnet.  Just don’t use it.  Problem solved!  Outta’ sight, outta’ mind!  The word lame comes to mind.  How, in this world of IP address space exhaustion, can you even begin to condone throwing away a perfectly good network for the sole purpose of preventing confusion with the network administrators?  Uh, you can’t.  Your network people need to get over it.  Suck it up, use the addresses.  “Waste is a thief!” (my token Fight Club reference)

But wait!  There’s more!

Second:  We didn’t just throw out the all zeroes subnet.  We threw out the all one’s subnet along with it.  The confusion surrounding the decimal value of the last network value being equal to the subnet mask and the hosts having a decimal value greater than the subnet mask combined with the whole, “Hey the broadcast address of the last network is the same as the broadcast address of the original network”, thing caused “they” (whoever they are) to toss out the all one’s subnet, too.  What was that word again?  Oh yeah!  Lame.

The all zeroes subnet and the all ones’ subnet are both perfectly valid networks.  They should be used and are used in environments that are movitvated to squeeze all of the usefulness possible out of the IPv4 address space.  So, if you haven’t been using it you need to get busy doing so.  Have I mentioned that NOT using subnet-zero in IPv6 isn’t going to be an option?  When your ISP gives you a prefix of 2001:ABCD:1234::/48 guess what your first usable network is going to be …2001:ABCD:1234::/64.  How do you like them apples?  Ha!

Colin Weaver