HSRP/FHRP in an OTV Network across DC's

[Bilal Nawaz]

Hello - I've been struggling to get my head around how to implement gateway redundancy over multiple DC's "hosting" the same L2 network(OTV). I'd appreciate if someone could lead me in the right direction to designing a solution.

What is the best way going about configuring HSRP in an OTV network across several data centers?

If I have a host in a DC1 trying to get to the HSRP standby address of 10.0.0.1 and the active member of this HSRP group is in DC2 how would that work? Will it traverse the OTV network to get to DC2 and then be routed back to DC1? or.... is there a mechanism that can be used for the .1 address to be active on the local core?

TIA,

Bilal

[Jerry Ye]

I am including a good document from CCO and this should be your reference:

http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9402/white_paper_c11-644634.html

This document will help you to understand how to localize outgoing traffic via the local DC's gateway.

For your question "Will it traverse the OTV network to get to DC2 and then be routed back to DC1?", it is really depends. In a normal case, yes, the traffic will routed to teh closest DC and L2 would switch over the OTV. In order to remediate this, there is a new protocol from Cisco called LISP. It will help optimize the L2 extended host return traffic.

http://lisp.cisco.com/

HTH,

jerry

[Bilal Nawaz]

Thank you for the documentation

So... the document uses a solution to segment and localise the HSRP  using VACLs and OTV mac control mechanisms, therefore primary core's in  either side assume the "active role". I guess that would work and  resiliency would be in place anyway as DC’s have both core's.

How will LISP work just for the purpose of providing local GW's?

[Jerry Ye]

LISP is not in used to provide local gateway. LISP is used to provide symatric traffic back to the L2 extended host(s).

For example:

Let's say Host-A is in DC1 VLAN100. VLAN100 is extended between DC1 and DC2. Host-A will take the default GW at DC1 to the outside world. Sounds fair?

Now, Host-A is talking to Host-B somewhere outside of the DC. Host-B see a route that it can go to VLAN100 (Host-A) via DC2. Return traffic is not asymmatric:

outbound - Host-A->DC1->CLOUD->Host-B

inbound - Host-B->CLOUD->DC2->OTV->DC1-Host-A

To make this symmatric, LISP can be used and you can read more about LISP from the link from my previous post.

Regards,

jerry

[mikemu]

Thank you Jerry you are my hero !