Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I had a smart (but non-technical) user ask me today:

Why did they pick 192.168.. for a private network address?

The only answer I could come up with is because 192 = 11000000 in binary. And 168 is 10101000 in binary. Both of which are kind of cool looking.

Is there a real historical reason for that particular choice of numbers? Why not 127.127..? Or 128.128..?

Similar question for 10.0.0.0 and 172.16.0.0

Thanks!

share|improve this question
    
the way you look at it, all binaries will look good :) 101000 1000100 1000010 –  Devrim Sep 10 '09 at 15:03

4 Answers 4

It appears they where selected by IANA simply because they where unused. The earliest mention I can find in an RFC is 1597. Also see rfc 1627. Both rfc have been obsoleted by rfc 1918

share|improve this answer

127.0.0.0/8 is a lookback subnet and thus cannot be used as a network spread around multiple machines. Similar applies to 128.x.x.x and other subnets which afaik are predefined by IANA to fall into the a, b, or c class of subnets based on their prefixes. This is probably the cause why 10.x.x.x, 192.168.x.x, etc are so wide spread - because they fall into different subnet classes. The numbers itself then were not chosen for a specific reason, they were simply unassigned or previously reserved as Zoredache already answered.

share|improve this answer

Before classless networks were invented, it was decided to make three ranges of private addressing space. These were:

  • Class A: 10/8 (the old ARPA reservation)
  • Class B: 172.16/12 (one of the first available class Bs)
  • Class C: 192.168/16 (one of the first available class Cs)

There are:

  • 1 Class A private prefixes (16.7 million addresses)
  • 16 Class B privates (65536 addresses each, totalling ~1 million addresses)
  • 256 Class C privates (256 addresses each, totalling 65536 addresses)

It is important to note that "Classes" haven't existed since 1994, and these days we use CIDR, which has a variable length subnet mask.

share|improve this answer

I forwarded the question to the internet-history mailing list and Craig Partridge, chief scientist at BBN, said:

10.0.0.0 is easy. For folks who needed LARGE private networks the only large space available by the early 1990s was the old ARPANET network number (the ARPANET was net 10 and was decommissioned around 1991).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.