HSRP or VRRP with subnet /30 ?

[sebastien3]

Hello,

I would like to replace an old router with a couple of new router in order to use HSRP or VRRP.

The router is currently connected to other routers so that each has a dedicated /30 interconnect subnet and dedicated Vlan.

It is not possible to be able to use HSRP or VRRP in /30 because an address is missing...

_R0_--10.0.1.2/30---SWITCH---10.0.1.1/30--R1

_R0_--10.0.1.6/30---SWITCH---10.0.1.5/30--R2

_R0_--10.0.1.10/30---SWITCH---10.0.1.9/30--RX

Is there a solution to use redundancy with /30 ?

Thank you !

[Richard Burts]

As you mention, when using a /30 there is room for only 2 host addresses. So redundancy using HSRP/VRRP is not possible for a subnet with /30 subnets.

Perhaps there is something in this environment that I am not understanding. But let me ask a fundamental question: what kind of redundancy are you looking for? What you appear to be describing is R0 with a single physical connection to R1, a single physical connection to R2, and a single physical connection to R3. What kind of redundancy can you achieve with a single physical connection? At layer 2 I do not see how redundancy is possible unless you can provision additional connections. At layer 3 redundancy could be possible if you provision a connection between R1 and R2 and a connection between R2 and R3. With the additional connections R0 would have 2 physical paths to each router, and running some dynamic routing protocol between the 4 routers could provide layer 3 redundancy.

[Joseph W. Doherty]

What I've seen done in situations similar to yours, shift all traffic on one of your old routers.  Replace the standby with it's replacement, still configured as standby.  Start to shift traffic to it, at whatever pace is believed "safe" (with verification all is working as expected). Then repeat replacement for other old router.  Lastly, reconfigure usage priorities as desired.

The forgoing, though, does during actual replacements leave you without physical redundancy, but if well planned, usually its not a long time window you're without redundancy.  Of course, ideally you make these changes during maintenance window(s) or at least as a lowest peak usage window.

The forgoing often also provides similar transition for a router's L3 links, otherwise adding an additional router creates additional issues, often more intractable than moving FHRP primaries from one device to another.

[sebastien3]

Hello @Richard Burts @Joseph W. Doherty 

Thank you for your answers. Here the R1/R2/RX routers have no connection between them... Routers is seprared.

It is not possible to go from a /30 to /29

The only solution I see is to use FlexLink on the switch in order to send the traffic to the router which will be the main one, in case of problem on the main router if the port goes to the DOWN state the Backup Flexlink will flow the traffic to the second router ...

[Joseph W. Doherty]

If replacement routers cannot be attached to same LAN side as existing routers, how would you expect FHRP to work, as it needs to share L2?

What I described doesn't require the replacement to actually physically replace the existing router, but it would require sharing the L2 LAN side.

[Richard Burts]

In my first response I asked this question " let me ask a fundamental question: what kind of redundancy are you looking for?" Do you have an answer for that question.

If R0 has a single connection to R1, a single connection to R2, and a single connection to R3 then how can there be redundancy?

Given the environment described in the original post (especially R0 with single connection to 3 routers and no other  connection between those routers) and the restriction that the subnet must be /30 I do not see a solution that provides redundancy. If redundancy is important then you need to make some changes in the environment.