2 questions regarding OSPF/BGP

[neilcampbell1]

We have a setup as described in the attached picture,  I was wondering if someone could answer a couple of questions about it,  I'm pretty new to routing but have a passing understanding.  But I can't work out:

1.  All inter Datacentre traffic currently traverses the LES.  It is now getting overloaded.    Can we route just  1 vlan (vlan 55 for instance) across the MPLS, without a static route, so we'd have automatic failback in the event of a loss of Datacentre router and no need to remove that static route?    If so would this be done by our ISP, or is it something we can control with OSPF?   (All SVI's for the vlans in the Datacentres exist within their respective cores)

2. We originally ran EIGRP internally and before we switched to OSPF,  our sites were split when searching the internet, half were forced to Datacentre 1, the other half to Datacentre 2.  In the event of a loss of Datacentre, the traffic would route the other way.  Since the switch to OSPF, all traffic arrives at Datacentre 2, and our ISP say that they have changed nothing.   Is this their responsibility, and is there anything we can do our end to change this using OSPF?

[Peter Koltl]

You cannot use a direct link between non-backbone OSPF areas for routed traffic.

What you might need is an OSPF sham link that could be an alternative for your direct interconnect link but you'd better consult with the ISP.

[neilcampbell1]

Hi, thanks for your answer!  Unfortunately consulting with our ISP is pretty hard,  question 2 up there is a bit of a testiment to their current attitude,  they control all of the routes coming into us but deny that they've changed anything. 

I'll have a look into sham links though,  thanks for your help!

[rizwanr74]

You can inject default-route into OSPF routing domain via both core switches to OSPF speakers.

Under the OSPF processor "default-information originate" in the event Datacentre-1 loses its own default-route, all other OSPF speakers will withdraw default-route learned from that specific OSPF neighbor and will use other default-router learned from Datacentre2, assuming both datacentres WAN circuit have equal cost.

router ospf 100
 default-information originate

[neilcampbell1]

Excellent, I'll start looking into that, so effectively I just add it as a static route in a route map, and if that route no longer exists, that will be ignored?    Perhaps I better read up some more on OSPF!

Cheers!

[rizwanr74]

You would not need route-map to advertise your default-route, just "default-information originate" will do the trick and if core switch looses its own static-default-route it will withdraw the static-route from advertising it to other OSPF speakers.

[neilcampbell1]

Hi,  we seem to be using this already:

router ospf 1

  router-id xxx.xxx.xxx.xxx

  network xxx.xxx.xxx.xxx area 0.0.0.1

  network xxx.xxx.xxx.xxx area 0.0.0.1

  network xxx.xxx.xxx.xxx area 0.0.0.1

  network xxx.xxx.xxx.xxx area 0.0.0.1

  default-information originate

  redistribute static route-map STATIC->OSPF

  area 0.0.0.1 authentication message-digest

  passive-interface default

along with this static route-map...

Excuse my lack of knowledge here, but if within this static route map, if a static route is defined, will that continue to route to that hop if it's available, or more importantly, not route if it isn't ?

[rizwanr74]

If static-route next-hop is not available it will not route to and that given route will not be advertise any longer.