Is there a FHRP that works between subnets?

[michaeleastongodwin]

I'm curious about the existence of an FHRP that works between subnets on the same LAN. (I know this technically wouldn't be the packets "first hop" anymore, but I'm more so talking about the concept of dynamically changing the packets route out of the LAN). I'm still a beginner in networking, but after some research I can't find one that does this and I'm curious to know why not. It seems like it would add yet another layer of redundancy to a network if you let's say we had a distribution layer with 2 different units, "DistroA" and "DistroB". Each was made up of 2 layer 3 switches running HSRP, and each of those had a connection to a router that left the LAN. DistroA's router was the primary connection to the WAN, but it goes down, wouldn't it be of some use to have a protocol that could trigger an automated response to to make DistroB's router the default gateway out of the Lan for DistroA end devices as well, while the troubleshooting took place on the faulty router? 

I know this is hypothetical, but I love talking about this stuff and would enjoy hearing anyones opinion, knowledge, or corrections on the matter. Thanks!

@ hsrp @ fhrp @networkdesign

[Joseph W. Doherty]

Possibly what you suggest is already covered by some FHRPs where the active gateway changes based on what happens upstream.

[Richard Pidcock]

Can you post a rough diagram of the topology you're describing and the point of failure described.  I think (if things are properly designed and connected) that FHRP's will provide the needed redundancy.

[David Ruess]

Hello,

What you are describing seems to be related to object tracking. The premise is yes FHRP cant tell when upstream connection fails. So you can create an "track" that references either a route or connectivity to a certain IP/network and if that tracked object fails then it can decrement the FHRP Priority. Then as long as you have preemption enabled the FHRP will take care of the rest.

Here is an example config:

conf t

track 33 ip route 192.168.1.0/24

interface vlan 10
standby 10 track 33 decrement 25

If the router loses connectivity to the 192.168.1.0/24 network it will trigger the track and decrement the priority of Standby group 10 by 25. You can also set it for line-protocol. So if the link goes down between the Switch and router it can also track that

Here is the documentation that explains all the things you can do with object tracking. You can track routes, interface status, set a delay to wait before it decrements the priority, track multiple objects

https://www.cisco.com/c/en/us/td/docs/switches/blades/3040/software/release/12-2_44_se/configuration/guide/swhsrp.html#wp1084266

Hope this help

-David

[Joseph W. Doherty]

@michaeleastongodwin what @David Ruess is describing is exactly what I had in mind.

Again, the gateway IP doesn't change, only the device that's actively processing for that IP changes (much like if the active interface or router failed, but without that kind of failure).

BTW, some FHRP variants can host multiple different subnet gateways on the same interface, but hosts wouldn't jump between subnets (unless dual homed).

Also BTW, since the introduction of stackable switches, and their variants, generally use their gateway redundancy vs. using a FHRP.

[paul driver]

Hello
Upstream from the layer3 core switches usually would be running an IGP/EGP that dynamically advertises prefixes/default routes for external traffic, These would be advertised to both L3 cores switches with preference given to specific path based on say a preferred advertised seed metric -any failure to the preferred upstream connection would result in dynamic failover to the less preferred exit point, this can be incorporated with some conditional monitoring(IPSLA/Tracking) for a more deterministic fail over