Subnetting for Dummies

Have a question or want to start a discussion? Post it! No Registration Necessary.  Now with pictures!

Threaded View
I'm starting trying to understand subnetting. Any help would be

Follow me here:
- Lets say I have one public class C address from my ISP
- Lets call it
- Lets say I have 500 clients.

Why do I even need to subnet? NAT is confusing me... Can I not just
have a router with NAT and then do whatever the heck I want on my
LAN?  (like my two home computers do).

Router WAN ~
Router LAN ~

In the above scenario I can just throw everyone on the 10.x.x.x
network and be done? Do I only need to subnet when issues of bandwidth/
collisions come up? I understand all about broadcast/collision
domains.. but I'm confused about how subnetting "saves address space"
when you can just use NAT to split one address into a million...

My head hurts... if anyone can point me in the right direction without
using any big words it would be much appreciated....

Re: Subnetting for Dummies

You cannot make a network with 500 clinets in class c address if you
want all clients to be in the same subnet, because class C can hold
253 clients. You have to left one address to represent the network
(, one for representing interface (example
and the last address int the subnet is  a broadcast addres
s( So if you want to make network for 500 clinets you
have to use class B address (example: If you
use this addres without the subnet you can have 255*255 clients in the
same subnet. If you want to preserve address space then you use subnet
and you use because with this subnet you have
2*255 and that is enough for your 500 clients.
 And if you have one public ip address you have to use subnet just to
make all clients in one subnet, bu you cannot use nat because nat can
make from one private address to one public. You have to use pat,and
all 500 clients can go out on the same address and the router will
know from which address it came, because router for every address use
different port.


Dennis je napisao/la:
Quoted text here. Click to load it

Re: Subnetting for Dummies
CrazyHorse wrote:

  >and you use because with this subnet you have
  >2*255 and that is enough for your 500 clients.

With the above netmask of (actually /22) you'll have 4*255-3
addresses. The desired 2*255-3 would be (/23)



Re: Subnetting for Dummies
Quoted text here. Click to load it

I think you misread? Dont confuse me more! :) In the above scenario,
my LAN network was using class A, mask

Quoted text here. Click to load it

I understand that. But I dont why would I need to "preserve address
space" on my LAN (using the above scenario)....? Im starting with a
class A address of are there not endless possibilities and
combinations of subnets I could use? This is where I am a bit foggy.

Quoted text here. Click to load it

Explain please? On my home lan my router uses NAT to split my one
public IP address from my ISP into countless local IP addresses. What
is this 1:1 relationship you speak of?

Quoted text here. Click to load it

Hmm... I'm afraid to ask.. but what is PAT???

Re: Subnetting for Dummies
Quoted text here. Click to load it

Well, you don't actually HAVE TO subnet and preserve address space.  But,
you should keep an eye on the future and the bigger picture.  Suppose your
company grows and you end up with 30 locations.  Now, you have 30 subnets
that are needed (plus wan links, etc...)  Well, okay, you can tap into the
172 and 192 address.  But, if you had the same short sighted attitude, they
would run out quickly as well.  Also, as you grow, you will want to consider
address summarization.  When you don't consider your address plan ahead of
time, this will become a cumbersome process and you will end up needed to
renumber at that time.  As they say, a stitch in time...

Yes, there are endless (well, virtually endless) possibilites using the 10
network.  But, you need the forsight to consider that ahead of time.  What
I've typically done is reserve one class C size address per location
(,, etc...).  If the site requires more address, you
can consider a larger address block (

Also, consider if you want different DMZ blocks within the Class C that the
ISP assigned.  You need to subnet that block into the appropriate number of
subnets/hosts and apply to a multiport firewall.

So, if you don't have the need at this point, there is nothing that will
force you to.  But, it is good practice.

Quoted text here. Click to load it

PAT is really what your home router is doing.  NAT translates one outside
address to one inside address, thus the 1:1 reference.  Your home router
will dynamically assign a port to you when you make an outside request, not
an address, since it only has one to share.  That is why it is Port Address

Hope that helps,


Re: Subnetting for Dummies

Quoted text here. Click to load it
Microsoft calls PAT "Internet Connection Sharnig".

Re: Subnetting for Dummies

Great posts guys thanks for info! Everything is starting to make sense

Re: Subnetting for Dummies

Dennis je napisao/la:
Quoted text here. Click to load it

Quoted text here. Click to load it

You dont't have to do it, it can be done without subnet, but if you
learn subnetting it will help you in the future,  when you wouuld be
dealing with large networks, and then you will have to be carefull.
But you're right, class A has many many possibilities and if you use
it for only one LAN, don't be bother with subnet :)

Quoted text here. Click to load it

PAT is NAT with overload. It means that it uses differnet ports for
every ip address in your LAN whet it translate private in public. With
PAT you can in theory translate 60000 ip addresses but the real number
is around 4 000

Site Timeline