Date: Sun, 02 May 1999 18:17:23 -0700 From: Greg Shenaut <greg@bogslab.ucdavis.edu> To: questions@freebsd.org Subject: variable-sized subnets in class C Message-ID: <199905030117.SAA16692@deal1.bogs.org>
next in thread | raw e-mail | index | archive | help
I am trying to divide up my class C network into variably sized
subnets. I think I understand how to do this, but I thought I'd
write up what I think to be the constraints, and see if anyone
more knowledgeable than I could keep me from making a mistake.
The rules are: the highest and lowest address in each network cannot
appear in a subnet. Always subnet in powers of two. When assigning
host IPs within a subnet, the lowest subnet address (all zeros) is
reserved for "this host", and the highest subnet address (all ones)
is reserved for the subnet broadcast address. Smallest useful
subnet is size 4; largest useful subnet is size 64. This is because
if you have two subnets of size 128, they will both be invalid
because of containing the lowest or highest network address; also,
subnets cannot be of size two, because both addresses would then
be reserved ("this host" or subnet broadcast). The lsbs of subnet
masks are equal to 256-s, where s is the total number of addresses
in the subnet (and from this it follows that the 0 bits in the
subnet mask will always be contiguous & on the right). Also, by
convention the highest usable address in a subnet is given to the
default gateway for the subnet, if there is one.
These are the largest legal subnets possible in an 8-bit class C
network:
0-3 -- 4 unusable addresses
4-7 -- 4 addresses (max 2 usable)
8-15 -- 8 addresses (max 6 usable)
16-31 -- 16 addresses (max 14 usable)
32-63 -- 32 addresses (max 30 usable)
64-127 -- 64 addresses (max 62 usable)
128-191 -- 64 addresses (max 62 usable)
192-223 -- 32 addresses (max 30 usable)
224-239 -- 16 addresses (max 14 usable)
240-247 -- 8 addresses (max 6 usable)
248-251 -- 4 addresses (max 2 usable)
252-255 -- 4 unusable addresses
All legal subnets can be subdivided recursively:
8 --> 4 + 4
16 --> 8 + 8
32 --> 16 + 16
64 --> 32 + 32
So for example, the range 32-63 could either have 14 valid IPs with
a subnet mask of .224, or be subdivided into two ranges, 32-47 and
48-63, with subnet masks of .240. Each of these ranges could either
have 6 valid IPs, or could be subdivided again into two subnets
with netmasks of .224, each with 2 valid IPs. So the 16-address
range can have the following possible subnet configurations:
1 14-host .224 subnet
2 6-host .240 subnets
1 6-host .240 subnet + 2 2-host .252 subnets
2 2-host .252 subnets + 1 6-host .240 subnets
4 2-host .252 subnets
The other subnets of size 8-64 can be subdivided according to the
same logic. What you can't do, though, is to combine any addresses
across binary tree boundaries. For example, you cannot have
* 1 10-host .??? subnet + 1 2-host .252 subnet
* 1 2-host .252 subnet + 1 6-host .??? subnet + 1 2-host .252 subnet
If you want, for example, to combine the two 64-address subnets
and one 32-address subnet to give one subnet with 160 addresses,
what you have to do is to set up the separate 64-address and
32-address subnets and route between them, for example by putting
three ethernet cards in a FreeBSD machine, one connected to each
subnet, and having it do the routing.
Well, if anyone spots a flaw in this, please let me know.
-Greg Shenaut
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199905030117.SAA16692>