DC to DC BGP loadbalance or active active

[satya mothukuri]

Hello,

we are planning to implement active active links between our 2 Data Centers. currently we have 1 link and we use BGP for connecting between the DC's. Now we got second circuit and want to make this as active active connection so that if the primary link(existing) link goes down, there should not be any down time between the Dc's. i have shared the network diagram. Please let me know how we can configure the routing? can we use existing GBP or we need to change that to EIGRP?.

[balaji.bandi]

i would suggest to use BGP where supported, so you have more Control on the TE compare to IGP.

below exmaple provide you can do many ways :

https://www.cisco.com/c/en/us/support/docs/ip/border-gateway-protocol-bgp/13762-40.html

[satya mothukuri]

Thanks for your reply.

I see many options only show load sharing and when there is a failure, we will see downtime for few sec or min. as there are much traffic between DC's, this may be an issue. we need 100% uptime between the DC's.

[balaji.bandi]

When you configured BGP, the BGP will take care of that portionof failure scenarios, you will not see any service interuption if you configured correctly.

You are not the first to deploy this kind of scenarios, world most of the network rely on BGP.

Since working solution changing most of the people mix the Links, since cost is very high for VPLS or p2p dark fibre.

So only reliable option for costing is MPLS.

[Georg Pauwen]

Hello,

BGP multipath load sharing is probably your best option. What ASs are the routers in ?

[satya mothukuri]

Running ebgp between the DC's.64787 and 64788

[MHM Cisco World]

that it not easy task, there are many many point must check. 

[Joseph W. Doherty]

You're attachment only shows the DC's connections to each other (correct?).  Much more information would be needed to recommend whether to use BGP or EIGRP in your case.

Either BGP or EIGRP should be able to use both links concurrently, and if one fails, shift all traffic to the remaining link (and, of course, if the fail link is restored, resume using it again).

Just using both links, and providing basic redundancy, shouldn't be overly complicated using either routing protocol.

Where routing configuration become more complex is when you want to minimize the impact (at time of failure) of a failed link.  I.e. you don't want some link failure "black holing" all the link's traffic for a minute or so.  Ideally, a link failure's impact would be so "light" active/live VoIP or video sessions would not notice, or barely notice, when it happens.

Also, BTW, to fully take advantage of using both your links concurrently, you might want to look into something like Cisco's PfR technology, which can dynamically load balance your links.