Ok, well, here goes from what I can remember.
 
127.0.0.1 - loopback address (Although the whole 127.0.0.0 - 127.255.255.255 range is really unusable)
 
10.0.0.0 - 10.255.255.255 - Class A non-routable
 
172.16.0.0 - 172.31.255.255 - Class B non-routable
 
192.168.0.0 - 192.168.255.255 - Class C non-routable
 
169.254.0.0 - 169.254.255.255 - Automatic Private IP Addressing (APIPA) Class B, and it's a Windows thing. You will be assigned an address in this range on Win2k and above DHCP clients that can't locate a DHCP server.
 
BTW, yes I gave the whole range and yes I know that the first and last addresses in the range are generally not usable.

Quote:172.16.0.0 - 172.31.255.255 - Class B non-routable

192.168.0.0 - 192.168.255.255 - Class C non-routable
Actually, if I remember my networking classes correctly:

192.168.0.0 - 192.168.255.255 is Class B, not class C (subnet mask of 255.255.0.0 - 16-bit host id)
172.16.0.0 - 172.31.255.255 is a subnet of Class A (subnet mask of 255.240.0.0 - 20-bit host id) (this is possible due to the introduction of subnet masks which effectively superceeded the original A,B,C class distinctions)

Quote:BTW, yes I gave the whole range and yes I know that the first and last addresses in the range are generally not usable.

Yes, in fact 0 and 255 have special meaning in an IP address. 255 means "broadcast". Sending a packet to 255.255.255.255 will send it to all IPs on the network (note that this is not forwarded by routers so don't think you can just send a packet to everyone on the internet ). Likewise sending a packet to 192.168.255.255 will send it to all IPs on the network that begin with 192.168.

0.0.0.0 means "this host on this net".

There are various other special combinations also.

s I understand it 169.x.x.x is fine. 169.254.x.x is reserved for autoassigning.
 
As for autoassigned addresses, it works for at least win98SE and above, not just win2k and above. (I used to have a network setup this way, long story)
 
Indeed 192.168.x.x mask 192.168.255.255 is class b.
 
class c? Who cares, its useless

Here is a link that should provide a bit more info:
 
http://www.cisco.com/univercd/cc/td/doc/product/aggr/vpn5000/5000sw/conce60x/5000cfg/ipaddr.htm
 
I got the numbers correct at least.
 
Not that it seems to matter much these days, but
 
Class A is 1.0.0.0 to 127.255.255.255 (10.x.x.x falls here)
Class B is 128.0.0.0 to 191.255.255.255 (172.16.x.x - 172.31.x.x falls here)
Class C is 192.0.0.0 to 222.255.255.255 (192.168.x.x falls here)
 
So my classes were correct.
 
I goofed, good call, APIPA does work for Win98 and above, wish I saved the link to that.
 
and an address like 192.168.0.0 where the host bits are all 0s identifies the network, while something like 192.168.0.255, where the host bits are all 1s is a broadcast address for that network.
 
Of course this will all be pointless whe IPv6 comes out and DNS will rule because God help you with trying to remember an IPv6 address.

] more lists that seem to have fallen through the cracks
 
224.0.0.0 through 239.255.255.255 --> used for multicasting as seen in the MBone of the internet (also known as class D)
 
240.0.0.0 and above is experimental (known as class e)
 
Just adding some $0.02

WHich IP address can be used on an internal network?
 
i am redoing our router and want to use something other then the 192.168 range.... if possible..

When we say not "outward" routable - that still means my router can use NAT to point to them for certain things internally? as in make my IP 200.8.8.8 on port 21 point to 10.10.10.10 and 200.8.8.8 port 22 point to 10.10.10.11 type thing.
 
 
this is all new to me!

Here is the slide set from my networking class. It provides a good explanation of how IP addresses work, and the ideas behind classless and classful addressing schemes.
 
http://www.ecse.rpi.edu/Homepages/shivkuma/teaching/fall2001/ccn2001-slides5a.pdf
 
AlecStar, these ip's (10.* 192.* etc) are considered non-routable because they are reserved for internal lan use. For example, lets say that you used 216.239.* ip ranges behind your router. If you wanted to go to google, its address (216.239.53.99) falls into the same class as your internal network and you would never get there because your router would believe 216... was an internal address. To prevent this, these "non-routable" ip addresses were set aside. By using them on your internal network, you can be guranteed that no host on the internet will have an ip address in your internal range.

I personally use the 192.168.xxx.xxx range as that is what my Linksys router uses for it's internal NAT/DHCP server.
 
Also note that when you subnet, it's really masking out certain octets and borrowing bits from others, hence why you would have issues connecting to one subnet when using another
 
This can get complicated unless you can remember to use the actual IP address rather then using NetBIOS and sharenames to connect via an intranet.
 
Having two separate networks can come in handy if you want to secure lower priority machines from say higher, more secure machines but then they can't share info if you need them too

Let's see what I remember from my CCNA classes... 8)
 
Internal, non-publicly-routable IP addresses are that way because the public routers on the internet block them. Check the RFC's for this standard (sorry, I don't remember which one...). These IP ranges are some of the first deny entries entered into the public routers. I purposely typed 'non-publicly-routable' above, because ANY IP address is routable. Again, the DENY statements in public routers are what keep the internal addresses from being routed across the internet.
 
The 'class' of an IP address has nothing to do with the subnet mask, but the location of the bits in the first octet. Example:
 
Class A: 0 --> 127
Bit Range: 00000000 -> 0xxxxxxx
Subnet Mask: 255.0.0.0
 
Class B: 128 --> 191
Bit Range: 10000000 -> 10xxxxxx
Subnet Mask: 255.255.0.0
 
Class C: 192 --> 223
Bit Range: 11000000 -> 110xxxxx
Subnet Mask: 255.255.255.0
 
Class D: 224 --> 239
Bit Range: 11100000 -> 1110xxxx
 
Class E: 240 --> 255
Bit Range: 11110000 -> 11111111
 
Notice how each class' bits are arranged. These are the rules! (Yet another RFC I can't seem to pull from my WOM...)
 
This does not mean you can't 'borrow' (subnet) or 'steal' (supernet) to make your range of usable IP addresses smaller or larger. However, the rules above still apply. Any IP range beginning with 220, for example, is still a Class C address.
 
Another $0.02 for the growing kitty...

Always, always, always consult the RFC's...
 
http://www.faqs.org/rfcs/rfc1918.html