02-10-2015 11:55 AM - edited 03-05-2019 12:46 AM
Hi there,
I have a number of Branches and ATMs which connect to the HQ via GRE tunnels through L2MPLS of the service provdiers network.
Recently I commisioned a DR site that I would like all the branches and ATMs to point to incase of disaster.
Most importantly I am supposed to achieve an auto-failover solution between Branches and ATMs towards HQ, @ATM and branch has duo links from different providers for resiliency.
The standard I am supposed to use is OSPF between branches and HQ, where we have GRE tunnels running in between, is there anyone who can assist me on how to achieve auto-failover solution between the Branches and HQ using OSPF on the existing GRE tunnels.
Sample configuration would really help
Thanks.
02-10-2015 09:53 PM
Hello Guys,
Any inputs on the above ?
02-11-2015 03:47 AM
What you are asking for here is a full blown network design. It is more than just a few configuration commands.
We can point you in the right direction but we cannot do the entire thing for you.
We would need to know things like is there a direct link between HQ and DR, how many branches, is OSPF already in use, if so what areas do you have, are you proposing to use the same IPs at the DR site etc etc.
But before all that have you thought about how the applications would work ?
Presumably you have applications that run on servers at HQ. How do you sync this information to the DR site servers ?
So a couple of scenarios -
1) the link at HQ fails and all sites automatically switch to DR. Then 10 minutes later the link comes back up so all sites switch back to HQ.
How are you going to make sure that any data written to servers in DR is now replicated to the HQ servers in real time.
2) a branch primary link fails. It switches to DR but all the other branches are still going to HQ.
Again how you are going to ensure the data remains consistent between the HQ and DR servers as you now have two active sites.
Routing protocols are very good at automatically providing failover but they don't understand the applications.
The hard part with DR is not the network, although that in itself can be challenging, but how the applications are going to work.
So if you only want to invoke DR if there is a major outage at your HQ sites which could last for days for example then using a dynamic routing protocol could create more problems than it would solve.
You may not have applications that need to be kept in sync so it may not be an issue for you.
But even then what you are asking for is not trivial, DR never is.
Perhaps you can clarify exactly how it is meant to work otherwise we cannot really point you in the right direction.
Jon
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide