Best Path Selection in ACI Fabric

Chella · ‎07-09-2016

In current network ,we have two boarder leaf switches and First Boarder Leaf SW is connected to CE1 and Second Boarder Leaf Switch is connected to CE2 .

L3_out is configured between boarder leaf and CE routers with OSPF. Each CE router is running eBGP with MPLS PE router and we have redistributed the BGP into OSPF and seeing all routes in ACI fabric .

From CE1 we have advertised the networks with network command and From CE2 also advertised same networks with AS_PATH prepend for backup because our requirement is to make CE1 link as primary and CE2 link as Backup for incoming and outgoing traffic.

I have two queries with respect to ACI fabric:

1. Tenant VRF is learning all routes from each CE router with default metric from both CE routers .query how ACI selects best path to forward traffic . In this case my observation is that always traffic exist from first Boarder leaf Switch.How is this happening ?

2. How to configure ACI such that traffic should always take the CE1 and when CE1 link fails then only CE2. What is the best option to do this?

Please suggest what is the best option to implement this .

dpita · ‎07-09-2016

Hello

The ACI routing table on your border leafs VRF is no different than any other routing table. it will select the best path based on AD, Metric, LPM etc.

From what i understood, you are peering to the ACI leafs from the CE routers with OSPF only? or is BL1->CE1 OSPF and BL2->CE2 using eBGP?

Chella · ‎07-09-2016

Hi Daniel,

It is OSPF between both BL and CE routers.

if I check on any other leaf switch ,it shows two routes as below , next-hop is different for different subnets.

10.2.0.0/16, ubest/mbest: 2/0
    *via 10.216.8.216%overlay-1, [200/1], 00:07:20, bgp-65100, internal, tag 65100
         recursive next hop: 10.216.8.216/32%overlay-1
    *via 10.216.8.217%overlay-1, [200/1], 00:07:20, bgp-65100, internal, tag 65100
         recursive next hop: 10.216.8.217/32%overlay-1

10.12.0.0/16, ubest/mbest: 2/0
    *via 10.216.8.217%overlay-1, [200/1], 5d08h, bgp-65100, internal, tag 65100
         recursive next hop: 10.216.8.217/32%overlay-1
    *via 10.216.8.216%overlay-1, [200/1], 4d00h, bgp-65100, internal, tag 65100
         recursive next hop: 10.216.8.216/32%overlay-1

Note :

10.216.8.216 - BL2

10.216.8.216 - BL1

I want all traffic should go to BL1. Can I modify OSPF metric while redistributing from BGP to OSPF or CAN I modify the BGP attributes while getting into ACI fabric. Which is simple method ?