Failover to DR site with BGP

Unanswered Question
Apr 20th, 2007
User Badges:

I have a question about a failover scenario.

We have an MPLS WAN with BGP as the routing protocol.

There is a web server at the main site and a failover web server at the DR site.

Say the HQ site web server were to die, would this be a DNS issue or can BGP be configured to route to the new Web serve?

Would the BGP be presenting our public IP addresses and directing traffic to which one is more desired?

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (2 ratings)
Danilo Dy Fri, 04/20/2007 - 09:03
User Badges:
  • Blue, 1500 points or more


First, both web server IP Address should be reachable thru internet BGP or no BGP (no BGP if you have static route to a single ISP). This is common nowadays.

Second, your DNS zone file TTL should not be higher than 1 hour (with high end compute nowadays this should be achievable). When the HQ site is down, change the "A" record for your web address from HQ IP Address to DR IP Address. Fro example;

from IN A

to IN A

If your web server have back-end application and/or database, make sure they are replicated/sync from HQ to DR site across private WAN and/or by Tape Media.

walleyewiz Fri, 04/20/2007 - 10:05
User Badges:

I agree completely. This is a DNS solution, not something for BGP. There are also some nice DNS solutions out there that will montior your website looking at the application and will automatically update DNS records in case of failure.

UltraDNS is one solution.

Wilson Samuel Fri, 04/20/2007 - 10:25
User Badges:
  • Gold, 750 points or more
  • Community Spotlight Award,

    Mobile User, July 2015


We have something exactly what is being discussed over here.

The easiest way to manage this is with the help of the Content Networking, and to be very specific Content Routing. Ofcourse our organization started when I guess Cisco's CSS gear were not very popular hence thus far we have been using quite effectively on F5s.

Implementing DR using the BGP for services would not turn in to a cost effective solutions as the Peering / Routing cost would boost very high!


Wilson Samuel


wilson_1234_2 Fri, 04/20/2007 - 12:21
User Badges:

According to the network team ( I have not been with this compnay very long), this is supposed to be handled by a CSS.

The CSS is supposed to redirect the traffic if the main site server is down to the dr web server.

How is this done?

acomiskey Fri, 04/20/2007 - 12:37
User Badges:
  • Green, 3000 points or more

Here is the document for configuring GSLB with DNS on a CSS. Basically, the CSS pair becomes the authoritative dns for The A records are located on the CSS's themselves. There is an app session between the 2 CSS's in which they exchange service, keepalive information etc. If the primary service is up, via a keepalive of some sort, the CSS will resolve to that VIP (A record). If the service is down, the CSS will resolve to the backup VIP. It works really fast.

wilson_1234_2 Fri, 04/20/2007 - 13:02
User Badges:


There should be two of the CSS devices, one in HQ, the othere in DR site?

acomiskey Fri, 04/20/2007 - 14:35
User Badges:
  • Green, 3000 points or more

Yes, in that scenario I believe they are sold in a pair as you need 2 for it to work.


This Discussion