I’ll pose and answer a few related questions along the way:
- Why do you see
255.255.255.0 so often? - Why
192.168.0.1? - Why
127.0.0.1?
Why such weird numbers — 255, 192, 168, 127?
8+8+8+8-bit dotted decimal
Internet addresses like 194.60.38.10 use dotted-decimal notation to split 32 bits up into 8+8+8+8 bits. Dotted-decimal means converting† each number to binary then left-padding it with 0’s.
For example .60. → 60=32+16+8+4 → 111100 → .00111100..
So 194.60.38.10 is dotted-decimal for the 4×8=32-bit address 11000010.00111100.00100110.00001010, since 38 → 100110, 10 → 1010, and so on. 194 requires all 8 bits; the rest are padded.
Once you think about 255, 192, and 127 in 8-bit binary, you can more easily understand why certain decimal numbers are so common:
- 255 =
11111111 - 192 =
11000000 - 127 =
_1111111 - 128 =
10000000
These decimal numbers happen to represent visually convenient 8-bit blocks like ■■■■■■■■, ■□□□□□□□, and □■■■■■■■. So you’ve never seen 256=2⁹ because of the 8-bit limit, and 127=128−1=2⁸−1 is the bit-flip of a power-of-two—and powers-of-two are 10………00000’s in binary.
Subnet masks: What’s mine is mine + What’s yours is yours
Subnet masks then break each 32-bit internet address up into a network ID and a host ID. Whereas internet addresses can have any mixture of 1’s and 0’s, subnet masks begin with only 1’s and end with only 0’s.
■■□□□□■□|□□■■■■□□|□□■□□■■□|□□□□■□■□ IP ■■■■■■■■|■■■■■■■■|■■■■■■■■|□□□□□□□□ subnet
Blacking out the first 8+8+8=24 bits and whiting out the final 8 bits is a way of splitting the IP ■■□□□□■□|□□■■■■□□|□□■□□■■□|□□□□■□■□ into two pieces:
■■□□□□■□|□□■■■■□□|□□■□□■■□ network □□□□■□■□ host
If the subnetwork owner (say OmniCorp) wanted more internal IP’s, they could buy up more (say 8+8=16 bits) of the righthand side of the network, like this:
■■□□□□■□|□□■■■■□□|□□■□□■■□|□□□□■□■□ IP ■■■■■■■■|■■■■■■■■|□□□□□□□□|□□□□□□□□ subnet ■■□□□□■□ □□■■■■□□ network □□■□□■■□ □□□□■□■□ host
Clearly there is a tradeoff within the 32-bit = 2³² = 4,294,967,296-option address space: if you buy up more network ID’s (lefthand side) your internal network has more host ID’s (righthand side) to assign.
Cheap people therefore have a subnet mask of 255.255.255.0 = ■■■■■■■■|■■■■■■■■|■■■■■■■■|□□□□□□□□. Even cheaper people have 255.255.255.128 = ■■■■■■■■|■■■■■■■■|■■■■■■■■|■□□□□□□□ or 255.255.255.192 = ■■■■■■■■|■■■■■■■■|■■■■■■■■|■■□□□□□□. According to folklore, it wasn’t actually Roger Miller, but a lowly sysadmin with a 255.255.255.254 mask who originally wrote King of the Road, substituting “I ain’t got a large subnet” for “I ain’t got no cigarettes”.
(Why are the masks of the lowly filled with such high numbers? Because, like Miller’s narrator, subnet masks count all the things you don’t have.)
What does the trailing slash after an IP mean? (eg, 194.60.38.10/24)
Since subnet masks (which divide “theirs” from “ours”) always begin with 1’s, and since we hate summing up powers-of-two even more than we hate figuring the powers-of-two in the first place, someone invented CIDR (the slash after an IP).
194.60.38.10/24 means "24 ones, then the rest zeroes in the submask", so ■■■■■■■■|■■■■■■■■|■■■■■■■■|□□□□□□□□ with 8+8+8 bits belonging to “them” and 8 bits belonging to “us”.
Reversing the hobo’s anthem above,
/31 is the songwriter/24 is middle-class (255.255.255.0 = ■■■■■■■■|■■■■■■■■|■■■■■■■■|□□□□□□□□/16 is rich ■■■■■■■■|■■■■■■■■|□□□□□□□□|□□□□□□□□/8 is super rich ■■■■■■■■|□□□□□□□□|□□□□□□□□|□□□□□□□□/1 or /0 would be the IANA or something.
† Use bc -l; obase=10; 60 for example.
