Static Default Routes

[aeroliteflyer]

Hello everyone,

I have a question some of you experience network designers or engineers may be able to assist me in.  My current dilemma deals with the static default route for my remote branches.  My architecture is basically a dual hub with spokes.  All branches come through the primary or then the DR site for internet access in an active/passive scenario.  Currently I have been controlling that but configuring two floating static default routes on each remote router.  There I use SLA/Tracking statements along with modifying admin distances to inject/prefer routes.  So recently I have been studying for CCDP and everywhere I read, including Cisco, states the preferred method in EIGRP to control paths is to modify the metric.  So, I have been testing using offset-lists with modifying the metric and that has been good so far.  Basically, on the private high speed WAN the DR has a slightly higher metric than HQ.  I also have a dual hub DMVPN as backups.  I have the offset-lists for those tunnels with a much higher metric than the WAN.  But by modifying the delay I still have active/passive paths for routes, with both tunnels at much higher metric than either WAN.  So, would you let EIGRP push the static default routes or configure static on remote routers? 

Thanks for any input,

Chris

[Richard Burts]

Chris

My preference would be to use EIGRP. If configured correctly it will react to changes in the network establishing the desired choice of primary or backup path through the network. It will do that automatically without the effort of running IP SLA or other route tracking that you currently need to do.

HTH

Rick

[Jon Marshall]

I don’t usually add to a post if it has been answered fully as Rick has here but just wanted to second Rick’s opinion here. 

In addition I would add as a general rule if you can run a dynamic routing protocol it is nearly always a better solution for the reasons Rick gives. 

Jon

[Richard Burts]

Jon

Thanks for the concurring opinion. I agree that if the network is small, and especially if for the most part there is only one way to get to destinations then static routing is a good choice. But as networks get larger, and especially when there are choices about how to get to destinations, and especially when you want to fail over to a backup path then dynamic routing protocols are generally the better choice.

HTH

Rick

[Joseph W. Doherty]

"I don’t usually add to a post if it has been answered fully as Rick . . ."

No, but I do, laugh, even when Rick's reply(s) are excellent, as they typically are, as is the case here too.

Generally, I second Jon's post (and Rick's second post), where EIGRP isn't mentioned, but just ". . . dynamic routing protocol . . ." I.e. there's nothing wrong using EIGRP, but just wanted to emphasis it's not the only dynamic routing protocol. I.e. besides the static routing vs. dynamic routing debate, you then move onto the proprietary vs. non-proprietary routing protocol debate.

I also wanted to mention, we don't know exactly how you're using SLA to trigger your static route changes. It's possible, your SLA configuration might react faster than a dynamic routing protocol would, especially one using default values. Some dynamic routing protocols can be "tuned", but perhaps not react the same as your SLA triggered static route changes.

[coolbreeze]

I agree with what has already been stated above as well - that EIGRP is the more desirable option.

Plus you have different options of how you can go about configuring the default route for advertisement through EIGRP.

For example, redistribute static default-routes directly or with a route-map, or configure a default-route summary address on an appropriate interface, or even using a distribute-list referencing ACL permitting default route.