Re: [Sheflug] Of networking, subnets and routing

["Alex Hudson" <nospam [at] york.ac.uk>]

> Following Richards ickle problem with his laptop, and not fully
understanding
> the routing table, folk may be intereted in this that was posted to the
> linuxchix techtalk mailing list a couple of days ago.

Um, firstly, 'linuxchix' ?! Have I missed some urban/hacker takeover of the
word 'chick' (other than poultry), or have you been hiding something Chris
?! ;))

> Someone couldn't
> understand their routing table and was advised (incorrectly) it was wrong.
> This bod, Malcolm, gave a good explanation of the routing table, and
netmasks.
> Hope it makes sense :)

I bet it's confused a load of people ;)) Technically correct, though, and
I've seen worse explanations. I guess it's another one of those reasons for
moving on to IPv6 - having to deal with with subnets that have masks
containing things other than "255" and "0" is a pain, especially if you need
to route between them. Also, it's easier to remember 58 as 32+16+8+2, since
one of my favourite mistakes is to always get the math wrong at the first
attempt (although usually a shift left or right one place fixes it - always
*one* power of two out ;) - working with binary will Do Your Head In.

As for why they choose 58 - it's not an amazingly special number, but does
have one useful property: you're unable to choose independent subnets which
share broadcast / network addresses. For example, if the magic number was 0,
the math would be easy, but all given subnets on that range would share a
network address, which is obvious silliness. The only possible way to do it
is to have network masks where the crucial fields are some power of two -
but then, you have a lot of silly networks which 'exist' but can host no
machines, so that's not really a drawback. Am I right in thinking this is
the lowest such number with this property? Actually, I don't think it is..
I've no idea how you'd prove something like that, though. I think the
property is no doubles appear more than once (i.e., at most one 00, at most
one 11), so 00101011 ought to work, and that's lower.. and 00010101 also
ought to work.. hmmm. Or am I completely off?

I think I've probably confused the issue here now..... oh well. For anyone
who's ever heard my explanation of mounting (Portakabins in Space, for those
who haven't :), this is probably no surprise.

Cheers,

Alex.

---------------------------------------------------------------------
Sheffield Linux User's Group - http://www.sheflug.co.uk
To unsubscribe from this list send mail to
- <sheflug-request [at] vuw.ac.nz> - with the word 
 "unsubscribe" in the body of the message. 

  GNU the choice of a complete generation.