Should I use Gre tunnels at my remote sites?

[birdsallc]

I have 40+ remote branch sites connected to the core via MPLS.  Those sites use bgp to the mpls.  Our internal network is 10.x.x.x/8.  The remote sites all use 10.x.x.x and this network is advertised via bgp to the rest of network.  I have eigrp as our IGP.   We redistribute bgp into eigrp and eigrp into bgp.

Remote site:

router bgp 319
 bgp log-neighbor-changes
 network 10.19.0.0 mask 255.255.0.0
 neighbor x.x.x.x remote-as 123
 neighbor x.x.x.x soft-reconfiguration inbound

Wan router:

router bgp 1
...
 network 10.0.0.0

...

Core switch:

router eigrp 99
 network 10.0.0.0

#show ip route | i 10.19.0.0

D EX    10.19.0.0/16 [170/3072] via 10.99.1.1, 2d05h, Vlan10

Question:  Should I use GRE (or mGRE) tunnels to my branches so that I can use eigrp for those subnets?

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Depends on your goals and needs.

Unclear what subnets you believe you need to advertise.  You're unable to get from some subnet to others due to EIGRP/BGP setup?

With GRE you'll likely need to deal with fragmentation, unless your MPLS vendor supports jumbo Ethernet, and you don't use it internally.

With an IGP, convergence is likely to be faster than BGP.

With EIGRP, you can also use unequal multi-path routing.

[birdsallc]

Joseph,   Everything still routes and traffic gets to it's destination.  Basically, it works as is.  I do noticed some strange routing behavior from time to time..  My main goal is to clean up the routing tables and have more efficient routing.

My personal preference is to keep the Lan and Wan separate and control what gets redistributed.  When I do a show ip route at my remote sites, it shows the long list of all of the routes via bgp.  I've always had the school of thought to keep the routing tables as small and as specific as possible.

So, I've got 10.x.x.x/8 advertised by both BGP and EIGRP.

WAN#sho run | b router eigrp
router eigrp 99
redistribute static
redistribute bgp 1 metric 442100000 1 255 1 1500
network 10.0.0.0
network 101.0.0.0
auto-summary
!
router bgp 1
no synchronization
bgp router-id 10.0.254.1
bgp log-neighbor-changes
bgp redistribute-internal
network 10.0.0.0

....

I guess I'm looking for insight as the best practice.  In my core, I see all of the bgp routes and all of 10.x.x.x/8 from remote sites as eigrp external.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

In general I completely agree to keep routing tables as small as possible.

You should be able to control what you redistribute between BGP and your IGP.

For example, sites might only need a default to get them to the BGP router.  The site's BGP router might only inject a summary for the site.

[Tagir Temirgaliyev]

you can do in one branch GRE and EIGRP and test it. here is remotly difficult to predict your goals and your design specifications. big routing tables not a problem now because routers now have a lot of memory and CPU.