HP3000-L Archives

May 1996, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Tom Emerson <[log in to unmask]>
Reply To:
Tom Emerson <[log in to unmask]>
Date:
Thu, 16 May 1996 12:07:00 +0000
Content-Type:
text/plain
Parts/Attachments:
text/plain (98 lines)
Mike asked:
 
Okay...I'm convinced that I do need a specific IP address on my HP3000. I
know that when my ASE was here back in November he used NMMGR to put in a
random IP address. Something like 192.0.0.0 or some such number. I guess
what I really need to know is does this number need to be a FCC licensed
IP address or can it be just a random number? I'm showing my ignorance
here on IP addressing.
 --
Mike Farrell
Intermountain Gas Company
Data Center Operations Supervisor
555 South Cole Road
Boise,  ID  83707
 ========================================================================
 
The [brief] answer:
   IF you are not planning on connecting the HP to "the internet"
      THEN use any random number, preferrably starting with 192.x.x.x (avoid
x.x.x.0, x.x.x.1, x.x.x.127, &  x.x.x.255)
   ELSE
      contact the NIC (Network Information Center) for a proper
domain/organization name and number.
   ENDIF
 
The [Longer] answer:
IP addresses are broken into 4 major categories, named class A, class B,
class C, and class D (actually, there is a class E and [I think] a class F
as well, but these are VERY rare -- you'll see why when I explain how these
work).  Think of IP addresses as a full 32 bit number broken down into two
components -- the network number and the computer number.  The "class"
defines how many bits are used to determine the network number, the
remaining bits are used to indicate which computer within that network you
are addressing.  To simplify matters, IP addresses are typically broken down
into four 8-bit quantities.  The class is derived by the number of bits used
at the head of the number (MSB's) and works out something like this:
 
   CLASS A: FIRST 8 bits
   CLASS B: first 16 bits
   CLASS C: first 24 bits
   CLASS D: first 28 bits [warning: shakey memory being accessed here...]
 
This means that there can ONLY be (theoretically) 256 class "A" licenses IN
THE WORLD, however, each "class A" license can encompass 2^24 computers.
 Likewise, there are 16386 class "B" licenses, each capable of addressing
16386 computers, and finally, (or, "most commonly"), class C licenses, of
which 2^24 can exist, but each can only addresss 256 computers.  Note that
these are "theoretical" maximums.  Since there needs to be some way to
differentiate the classes, there is a further breakdown within these bit
groupings.
 
For instance, given the above, would the address 1.2.3.4 be:
   computer 2.3.4 in class A network 1?
   computer 3.4 in class B network 1.2?
   computer 4 in class C network 1.2.3?
 
To resolve this ambiguity, the first byte of the address is broken down even
further to indicate which class the rest of the number is to be interpreted
as.  This breaks down to:
   CLASS A is assigned a range of 0 - 127
   CLASS B is assigned a range of 128 - 191
   CLASS C is assinged a range of 192 - 223
   CLASS D is assigned a range of 224 - 239
 
Or, if you look at the bit patterns, these are:
   00000000 - 01111111  (class A)
   10000000 - 10111111  (class B)
   11000000 - 11011111  (class C)
   11100000 - 11101111  (class D)
   and so on.
 
This effectively limits class A licenses to 128, class B to 64 * 256, and
class C to 32 * 16386.  Classes D, E, and F are used so rarely because while
there may be a great number of them, they can only access a few machines
within the network (in other words, if the first 28 bits of a 32 bit number
are being used for the network number, there can only be 16 machines on the
network, but as "machines" with numbers of all zeroes and all 1's are
pre-defined for other purposes, there can only be 14 local addresses...)
 
However, as I mentioned above, if you don't intend to connect this "network"
to "the internet", you can pick any scheme you want (although using class
"C" is recommended).  If you DO intend to connect to "the internet"
(eventually), then you must contact not the FCC (although that's a good
guess), but Internic, or Internet Network Information Center (usually just
referred to as "NIC") as they are "in charge" of administering class
licenses and network numbers.  This used to be a "free" service, but lately
I believe they have begun to charge $50 or $100 per year to maintain the
license (this is simply a supply/demand problem -- as I pointed out above,
the supply is certainly limited, but the "demand" is quickly approaching the
total number of computer users in the world as each and every user seems to
want "their own" internet address...)
 
There are plenty of books on TCP/IP that explain all the above in even
greater detail (including how to contact the NIC) which is where I'll direct
you to for further enlightenment.
 
Tom Emerson

ATOM RSS1 RSS2