nanog mailing list archives

Re: /19 addresses and redundancy


From: Al Roethlisberger <aroethli () cisco com>
Date: Wed, 05 Nov 1997 18:32:03 -0800

At 08:55 PM 11/5/97 -0500, you wrote:
Greetings,

Here is a question -- maybe someone out there can suggest a solution to a
problem...

We are a small to mid-sized ISP and we feel that in order to compete in
todays market we need to provide multiple circuits to different backbone
providers and run BGP.  We currently have a single T1 (from Sprintlink) 
which is handling our leased line, web, and dialup customers with little
trouble (running at about 20-25%).  We want to add another T1 to another
major carrier so we can offer a backup circuit in the event our T1 goes
down or if Sprint has a major outage.  Of course, this wouldn't be just
for backup -- it would share the load... 

We currently have 15 class C addresses and we have been told by Sprint
that anthing smaller than a /19 BGP may get filtered.  Since we have a
whole bunch of non-contiguous class C address (which Sprint gave us), it
seems that BGP would never work.

So, we contacted the Internic and requested a /19 block so we can do this. 
The Internic refuses to give someone a /19 block unless they are in need
of it right away (a few months projections).  We don't need 32 Class Cs
because we need to assign that many but we need them because we want to be
able to use BGP.  The Internic will not give them to us and we seem to
have no options. 


Are there ways around this?

Thanks,

Steve



Well, since Sprint is one of the few major ISPs really filtering at the /19
boundary for some addresses(see below), and since this IS their addressing;
you will probably do alright to try to advertise the networks via BGP.
Although I would recommend persuing your own portable addressing, there is
no reason that you should not be allowed to build some resiliency into your
network just because of the networking.

Although some networks have had some issues as times with longer than /19
prefixes, generally you can see from reports such as the CIDR report that
there are plenty of networks out there, in service, that are not being filtered.

Below is what is posted on the NANOG site as far as Sprint's filtering policy:

n the classical "A" space: accept nothing longer than /8      
          In the classical "B" space: accept nothing longer than /16
                  in 24/8 space: accept nothing longer than /19       
                  in 195/8: accept nothing longer than /19
                  in 206/8 - 223/8: accept nothing longer than /19
                  in 192/8 - 205/8: accept nothing longer than /24    
  
Agis also has a policy statement:

We will filter TWD at /24. 
     We will filter 206+ at /19. 
     We will filter everything else at /16. 
     We will apply next-hop-self to all exchange point peering sessions. 
     We expect peers to do likewise and we will route map if necessary. 


Again, whether these are adhered to strictly by these two providers, or any
others, will remain to be seen when you give it a try.

Keep persuing your own addressing in the meantime.

al




Current thread: