Two A records

Matt Prigge mprigge at pobox.com
Mon Apr 30 21:34:30 UTC 2001


What you really should be doing is running BGP with your ISPs. That way, if
one link dies, traffic automatically finds its way over the other link.
Incoming and outgoing traffic is also automatically load-balanced (in
theory). You also dont need to use two separate blocks of IPs that way.
Running BGP may require better router hardware and/or extra charges from
your ISP, but if you are really trying to obtain incoming and outgoing
redundancy, you need to do it this way.

To actually answer your question: to my knowledge there is no such thing as
a backup A record. Regardless, if your nameservers are on-site, youll still
need to have some way of telling the root-servers where to find your DNS
server (which is highly impractical). Even if your nameservers are offsite,
your only solution is to run with a very low TTL and reload a separate set
of zone files which point to the IP block served by your backup connection.
This is not only very kludgy, but also doesnt really take advantage of
having two connections to start with.

BGP is definitely the way to go.

----- Original Message -----
From: "Jennifer" <jkb at iccis.net>
Newsgroups: comp.protocols.dns.bind
To: <comp-protocols-dns-bind at moderators.isc.org>
Sent: Monday, April 30, 2001 2:14 PM
Subject: Two A records


> Is it possible to have two A records for one host? (not really one host,
> they would point to two different interfaces on the firewall.)
>
> I'm trying to set up a redundant T1 into my site from a different ISP.
I'm
> thinking that I won't have a problem load balancing the outbound traffic
but
> when a link goes down, The DNS will either have to update pretty quick, or
> I'll need a second record.  I could put two interfaces in my firewall, one
> on each subnet.  Then I could redirect port 80, 25, etc. to internal hosts
> from both interfaces.  I don't see a problem with the MX records because I
> can use a secondary.  Are there secondary A records?
>
> thanks for any help.
>
>
>
>
>



More information about the bind-users mailing list