[f-nsp] Internal vs External routing on an MLX

[Eldon Koyle]

HSRP is a Cisco-proprietary protocol, so your routers won't support it.
They do support the similar (but incompatible) VRRP.

Whether VRRP would work depends a little bit on how you will be
connected to the ISP.  If both of your MLXes will be connected to some
sort of switch which is connected to both of their redundant routers,
this would work the way you described.  Basically, all of the devices
involved with VRRP or HSRP would need to be on the same L2 domain.
Their cisco devices can talk HSRP, but you will also have to set up VRRP
on your end (otherwise, if the router with the IP address they are
routing to goes down, you still lose your internet connection).

Spanning a VLAN between your two MLXes would work if there isn't a
switch in the middle somewhere.  This could have unexpected behavior if
the link between the two MLXes were ever to fail, however.

HSRP/VRRP makes more sense when you want to have more than one router
able to take over being the default gateway for a switched network.  It
seems kind of icky to me to use it on a routed link (that's just my
uninformed opinion, though).

I would personally prefer to run BGP with your MLXes directly connected
to the provider's two routers as it has the least amount of stuff in
between that could go wrong.  It is possible that your ISP cannot
support that (if you are connected to a device that doesn't support BGP,
for instance).

-- 
Eldon Koyle
Information Technology
Utah State University

On  Aug 12  1:50+0000, Jimmy Stewpot wrote:
> Hello,
> 
> We currently have an internal backbone with redundancy.. we use private AS full mesh eBGP internally which has been working really well for a very long time. The problem that I have is we are about to move to a redundant inernet connection which can only be povided using HSRP by our provider. (our network has been colocated for some time.. we are just turning on an interet connection to our second MLX).
> 
> I am having trouble visualising what that will look like so that we can actually leverage the dual connections in a fail-over scenario.. The initial thought was to have a vlan that spans both "external" interfaces with a static route on both devices pointing to the destination.. my thought was that it would only have the MAC for the HSRP gateway on the active connection so outbound traffic would go across the inter-connection and out to the system.. Having never used HRSP I am hoping that someone can give me some light on what that topology woudl look like and if it would work.. also is it "best practise"?
> 
> 
> Any advice would be much appreciated.
> 
> Regards,
> 
> Jimmy.
> 
> 
> 
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>