IP addressing ... Im really dumb

[aimeeandbeatles]

I dont understand all this stuff about octects and network hosts and broadcasts. Please help?

 

[Turner]

Those are pretty broad topics there. Maybe something a little more specific?

 

[aimeeandbeatles]

Well the sheet gave a lot of stuff about broadcasting address and binary math and classed A and B IPs (???) and subnetting (I know this is done to manage networks) and public and private. It was very much stuff

 

[Cutlass]

Did they give you a book on it? You really need to study that as best you can. And only then post specific questions. Or find a website or check the library for a book. .

It's really too much to explain in a forum post.

 

[Till]

No need for books, this is basic stuff and Wikipedia covers it perfectly well.

Octect = 8bits = 1byte.

On IP classes and subnets (and the significance of octets):
http://en.wikipedia.org/wiki/IP_address#IPv4_subnetting

On public and private address ranges:
http://en.wikipedia.org/wiki/IP_address#IPv4_private_addresses

On broadcast addresses:
http://en.wikipedia.org/wiki/Broadcast_address

Please read the articles before asking more questions. If nothing else, they should at least allow you to make the specific enough to allow us to answer them without having to write essays.

 

[aimeeandbeatles]

Thank you ... The teacher was going very! fast! so Ill see if I can figure them out from the articles

 

[Turner]

And you can't be afraid to ask questions when you don't understand something. You might need to learn to do it outside of class, but do learn to ask when you don't understand.

 

[ParadigmShifter]

Tell them to slow down a bit perhaps.

Do you get printed notes or do you have to copy things down? Get copies of the notes in that case. Do they read out what they are writing? - in which case might be worth getting a voice recorder of some sort.

 

[innonimatu]

Well the sheet gave a lot of stuff about broadcasting address and binary math and classed A and B IPs (???) and subnetting (I know this is done to manage networks) and public and private. It was very much stuff

Sounds like a terrible teacher. I could understand using classes as an introduction (they're deprecated, but are easier to understand), but he was going to explain subnetting he might as well do it correctly.
And no, Till, wikipedia doesn't really try to explain this carefully. I guess that it probably merits at attempt at explaining it here:

The basics:
You just have to know that those user-friendly 1-255 numbers you see in the four parts of an IP address are really representations of a binary number. Each of the four numbers represents an "octet", a sequence of 8 bits.
2x2x ..., or 2^8, gives 256 combination, thus represented as 0...255. For ease of writing these 8 zeros and ones are represented as this single number. An IP adreess has four of these octets.

There are several "kinds" of IP addresses:
- a single host address, for example 192.168.57.25
- a network address, for example 192.168.0.0/16
- a broadcast address, for example, 192.168.255.255
- etc (the ones above cover the basics).

The way you convert a network address to binary is to take each of the four numbers and use some calculator to covert it, one octet at a time (I just can't be bothered to do it on my head):
So 192.168.57.25 is 11000000.10101000.00111001.00011001
and 192.168.0.0/16 is 11000000.10101000.00000000.00000000
and 192.168.255.255 is 11000000.10101000.11111111.11111111

As you can see a network address is just like a host address, except that the last bits are all zero. The network bits are the number of bits given in the "/x" notation, in this case 16. The first 16 bits. The rest of the bits are set as zero.
And the broadcast address is just the same bits as the network bits, but with the rest all as ones instead of zeroes.

Some obnoxious bastards insist in representing a network address not as 192.168.0.0/16 but as:
192.168.0.0 netmask 255.255.0.0
in other words, a "netmask" is just the network address with all its network bits set to one. This is easy to calculate when subnets match an octet border (/8, /16, /24), but the broadcast address (and the broadcast too) becomes a PITA to calculate when they don't.

Subnetting:
Now, it's hard to explain this rationally without explaining how routing protocols work. I could just say "it's like this", but you wouldn't understand why the "bits" are important. So:
The IP address is, well, the address for each packet of information. When a packet reaches one of the (possibly) many routers on its way, the router compares the destination address (typically, a host address) on the packet with the network addresses of its own networks. It does so by matching the initial bits of the address to the bits of each network, up to the number of bits of the network given in the "/x" in the end of the network address.

Consider a router which did had the 192.168.0.0/16 network and received a packed meant for host 192.168.57.25. Because it's a router, it also has a few other networks.
Suppose it will be comparing, in binary:
host - 11000000.10101000.00111001.00011001
network1 - 11000000.10100000.00000000.00000000 (/12)
network2 - 11000000.10101000.00000000.00000000 (/16)
network3 - 11000000.10101000.11100000.00000000 (/24)

Without getting into routing, because it's not necessary, I'll just add that routers will route packets into the most specific network, i.e., that which matches the most bits of the destination address with the "/x" bits of the network address, and only those bits. So, dumping the padding zeroes, we have:
host - 11000000.10101000.00111001.00011001
network1 - 11000000.1010 (/12)
network2 - 11000000.10101000 (/16)
network3 - 11000011.10101000.11100000 (/24)

So for the example above the router will compare the first 12 bits of the host address with network1, and the first 16 with network2, and the first 24 with network3.
The first and the second comparison match, so this "tie" will be broken in favor of the most specific: network2.

Now, why do both network match? What does that mean? Well, it means that network2, in the example above, is logically a subnet of network1. You wouldn't know from looking at its representation in decimal:
network1 - 192.160.0.0 (/12)
network2 - 192.168.0.0 (/16)
but in binary you can see it:
network1 - 11000000.10100000.00000000.00000000 (/12)
network2 - 11000000.10101000.00000000.00000000 (/16)

(and yes, it is valid for two networks to be connected to the same router even when one is a subnet of the other. The rules make sure that there's no ambiguity in the routing of traffic)

The size of a network address, as you can see, can be chosen arbitrarily. It was not always so: initially there were only supposed to be networks with three sizes: /8, /16, /24, the old classes A, B and C, but that's all ancient internet history. It was easier, though!

Now, with subnetting, you can have a network which is, for example, a /12 like the one above, given to you as 192.160.0.0/12, and in order to calculate the broadcast address you have to convert it to binary and then replace all the extra zeros with ones.
Starting with:
192.160.0.0/12,
convert to binary:
network1 - 11000000.10100000.00000000.00000000 (/12)
replace the last (32 - 12 = 20) zeros with ones:
network1 broadcast address - 11000000.10101111.11111111.11111111 (/12)
and convert back to decimal:
network1 broadcast address - 192.175.255.255

Not at all intuitive, unlike the old class system. But it's what we must use now. But don't complain too much, we're supposed to move to IPv6, and then the PITA will be worse!

 

[aimeeandbeatles]

That makes a bit more sense, thanks