RE: Reliability of a /24

["Wilfried Woeber, UniVie/ACOnet" <]

  Tim,

>In 1995, I was assigned a Class C network.  In 1995 I used it, but in
>1998 my organization moved out of state.  When we set up our Internet
>connection at the new location, we just moved forward with using the
>addresses given to us by our ISP.

  from the context in the rest of your message, I guess the "Class C
  network" is from TWD (toxic waste dump, i.e. 192/8, "the swamp"),
  correct?  Also, I assume that you are "based" in Europe (whatever that
  measn in NetLand :-).

>Today, our ISP doles out IP addresses out with a teaspoon and charges
>fairly serious money for more.  We would like to begin to use our /24
>network instead of dealing with getting more addresses from our ISP.

  ISPs are supposed to only assign that amount of addresses which are
  necessary for your application, and for which doumentation has been
  provided.
  While IPS are free to charge for the administrative overhaead in doing
  so (and of course for accepting prefixes for the routing layer), they
  are not supposed to charge for the addresses per se.

>In addition, we want to connect our network to a second ISP and begin
>broadcasting our routes via BGP.  The problem is, with the change to
>CIDR, it seems that /24 networks are unappreciated.  ARIN states that
>any network smaller than a /20 is not guaranteed to be globally routed.
> Yet other sources seem to suggest that most (but not all) /24's should
>be OK.

  This is mixing different aspects.

  First of all, the Regional Internet Registries (RIRs) themself do not
  have a *direct* influence on routing decisions taken by ISPs, and thus
  are not at all in a position to "guarantee" anything on the routing
  layer.

  However, the 3 RIRs do coordinate the address distribution framework
  *for those addresses* which are obtained from *their* blocks, and they
  do document the structure/size of those address blocks from which
  sections are allocated to LIRs/IPS.

  For quite a long time, those addresses were taken from blocks of size
  /19, with a recent trend to migrate to a (smaller) block-size of /20.
  This in turn allows the ISPs to configure routing filters *for those
  address ranges*. So, within that context, ARIN's statement does make
  sense.
  
  On the other hand, again assuming that your /24 is from the 192 swamp,
  those rules do not apply directly, because those addresses were
  distributed "all over the places". This in turn has motivated most ISPs
  to apply different route filtering approaches for those addresses, if
  at all. Again, in that context, a /24 from 192/8 is probably "OK".

>We don't want to switch to the /24 if that is going to give us routing
>problems.  In my search for information regarding this, I came across
>your e-mail address on a Web archive of a mailing list.  You seemed to
>have an understanding of this issue.  Could you please point me to a
>somewhat more difinitive source of information regarding this I would
>greatly appreciate it.

  Before deploying the /24 (provider independent) address space, you
  should check with *all* of your presumable ISPs/up-streams under what
  (technical / financial) conditions they would accept and propagate (or
  source) your route announcements to their peers.

  Again, this is nothing for the RIRs to influence or to decide.
  
  From the point of view of CIDR (route aggregation) you might see some
  prodding to use ISP-provided (aggregateable) addresses....
  
  Hth,
  Wilfried.
 _________________________________:_____________________________________
  Wilfried Woeber                 : e-mail: Woeber@localhost
  UniVie Computer Center - ACOnet : Tel: +43 1 4277 - 140 33
  Universitaetsstrasse 7          : Fax: +43 1 4277 - 9 140
  A-1010 Vienna, Austria, Europe  : RIPE-DB: WW144, PGP keyID 0xF0ACB369
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  First things first, but not necessarily in that order....