border node redundancy

[rkdgus2344]

To introduce SDA, we are designing.
I am contemplating a design that considers border node redundancy.
When deploying border nodes with each single device, we worry that it will become a single point of failure when all external connections on one border are lost.
When composing border nodes into stacks, it seems that it is relatively easy to secure border node redundancy. If you have experience doing this, please help.

[Flavio Miranda]

Hi

 Please, take a look in the document attached. I believe this is going to help you a lot.

And I´d like to add that you are right. On this new concept of build networks based on software, you need to have redundancy on the underlay network. That´s why  is important to choose platform  that allows you work with stack ultimately. This way you can simplify your design by "logicaly" having one device but, at the same time, be protected by hardware failure with stack. 

[markus.forrer]

If it's a new deployment probably the new Lisp Pub/Sub with default dynamic Border is something to look at.

If BGP Handoff is used the you need to have iBGP configured between the two Borders.

If you're prefer to have a stack, keep in mind that the SW upgrades have impact on the connectivity.