Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Community Member

Redundancy between buildings

Hi guys,

I have a situation here and I am not sure how do I solve it.

At one site I have an MPLS single link with subnet 192.168.1.0/24 connected to L3 switches.

At the other site I have another MPLS single link with subnet 192.168.2.0/24 connected to L3 switches.

These should be configured Active/Passive , like all MPLS traffic should be flowing through site1 as long as that is up and thru site 2 if site 1 goes down.

These sites are connected with 2 P2P links .

The problem I have , is that the requirement is to have a sort of DR in place for these buidings in such fashion , that if site1 MPLS goes down all traffic should flow through the site2 MPLS link.

I have attached a picture , if that would help!

Thanks

VladMPLS_HA.jpg

1 ACCEPTED SOLUTION

Accepted Solutions
Hall of Fame Super Blue

Re: Redundancy between buildings

Vlad

Okay, if you are not exchanging routes with the ISP then you can do this. Lets assume site 1 is active and site 2 is passive in normal operations,.This also assumes the the 2 MPLS routers can route to each other via EIGRP.

1) Configure a static default-route on site 1 router that connects to MPLS pointing to the next-hop for the site 1 MPLS connection.

2) redistribute this static into EIGRP. Note this will mean the static has an AD of 170 on all other routers in both sites

3) Configure IP SLA on the site 1 router to track the availability of the MPLS link. You need this because you are not exchanging routes with your provider so you won't automatically know if the link has failed

4) On site 2 router that connects to MPLS  configure a floating static default-route pointing to the next-hop for the site 2 MPLS link. Note the AD of this route must be > 170

5) On the same site 2 router redistribute static into EIGRP.

This works as follows -

1) The site 1 router connecting to MPLS redistributes the static into EIGRP. All routers in both sites see this route and they will all see it with an AD of 170.

2) The site 2 router also sees this route and because it has a lower AD than it's own static route then it uses this one. It will not insert it's own static route into the routing table and hence into EIGRP because it's own static route has a higher AD.

3) Under normal conditions all traffic follows the default-route out via site 1.

4) If the link or the remote provider router fails then site 1 router removes the default-route from it's routing table. It no longer gets redistributed into EIGRP and so site 2 router now no longer receives it.

5) site 2 router now installs it's default-route and redistributes into EIGRP and all routers receive this route. Note all routers except site 2 will see this route with an AD of 170.

6) If the link comes back up at site 1 IP SLA will reinstall the removed static route and then this will get propogated to site 2 router. Site 2 router will then see it's own default route with an AD > 170 and the new one with an AD of 170 and will use the new one. So it removes it's own default route from the routing table and stops redistibuting into EIGRP.

7) All intermediate routers will now use site 1 again.

Note - when site 1 comes back up and installs it's own route and redistributes every intermediate router will have 2 default-routes in their routing table temporarily until the site 2 MPLS router receives the site 1 redistributed route. Once it does and it removes it's own default-route that will no longer be redistributed into EIGRP and so all the intermediate routers will only have one default-route left which will be via site 1.

And it would need testing but i have used this sort of setup before and it works fine.

Jon

8 REPLIES
Silver

Re: Redundancy between buildings

i can't understand your problem. can you explain more?

where r these MPLS connections in the diagram? i suppose the connections you draw are the leased lines?

Community Member

Re: Redundancy between buildings

Those clouds are the MPLS leased lines....

If building one LL goes down , the traffic should be switched thru the B2 LL.

Basically , this should go like this:

    In a normal operation all traffic into that WAN cloud should flow from B2 thru the P2P links to B1 and out into the provider's cloud.

In case B1 LL is down , the traffic should flow from B1 thru the P2P to B2 and out .

The P2P will carry traffic only at L3 , no vlans , no L2 traffic .

What my actual problem is , how do I carry the HSRP or failover information from one building to the other?

Thanks,

Vlad

Hall of Fame Super Blue

Re: Redundancy between buildings

Hunnetvl01 wrote:

Those clouds are the MPLS leased lines....

If building one LL goes down , the traffic should be switched thru the B2 LL.

Basically , this should go like this:

    In a normal operation all traffic into that WAN cloud should flow from B2 thru the P2P links to B1 and out into the provider's cloud.

In case B1 LL is down , the traffic should flow from B1 thru the P2P to B2 and out .

The P2P will carry traffic only at L3 , no vlans , no L2 traffic .

What my actual problem is , how do I carry the HSRP or failover information from one building to the other?

Thanks,

Vlad


Vlad

Need some more details concerning routing ie.

1) how are you peering with MPLS provider - are you running BGP ?

2) what, if any, internal routing protocol are you using - EIGRP/OSPF ?

3) Are you redistributing anywhere

4) are the routers connecting the P2P links peering with each other ie. across sites.

5) You mentioned HSRP - where is that running. Are the switches in the diagram L3 and doing inter-vlan routing ?

Jon

Community Member

Re: Redundancy between buildings

Jon ,

There is nothing set up yet , this is a design this customer wants.

The issue they have is that the provider does not want to customize their routers' config to BGP with them on one hand.

1) how are you peering with MPLS provider - are you running BGP ?

   NO

2) what, if any, internal routing protocol are you using - EIGRP/OSPF ?

EIGRP , but OSPF is not excluded between the buildings and those core switches and redistribute static .

4) are the routers connecting the P2P links peering with each other ie. across sites.

Yes they will , and they will run an internal Routing protocol .

5) You mentioned HSRP - where is that running. Are the switches in the diagram L3 and doing inter-vlan routing ?

There is no HSRP as of now. The L3 switches are doing routing and the SVIs are the def gateway of teh PCs.They would then route into the LL and between these buildings as required.

This is the issue I have , since the ISP does not want to change any of their config on their managed WAN routers.

Vlad

Hall of Fame Super Blue

Re: Redundancy between buildings

Hunnetvl01 wrote:

Jon ,

There is nothing set up yet , this is a design this customer wants.

The issue they have is that the provider does not want to customize their routers' config to BGP with them on one hand.

1) how are you peering with MPLS provider - are you running BGP ?

   NO

2) what, if any, internal routing protocol are you using - EIGRP/OSPF ?

EIGRP , but OSPF is not excluded between the buildings and those core switches and redistribute static .

4) are the routers connecting the P2P links peering with each other ie. across sites.

Yes they will , and they will run an internal Routing protocol .

5) You mentioned HSRP - where is that running. Are the switches in the diagram L3 and doing inter-vlan routing ?

There is no HSRP as of now. The L3 switches are doing routing and the SVIs are the def gateway of teh PCs.They would then route into the LL and between these buildings as required.

This is the issue I have , since the ISP does not want to change any of their config on their managed WAN routers.

Vlad

Vlad

1) So how do you route to the MPLS network - do you have default-route somewhere ?

2) Not sure what you mean. Do you mean you are not running OSPF at the moment ?

Jon

Community Member

Re: Redundancy between buildings

There is nothing setup yet .

This is what I am tryin to figure out , as in how will we be able to do this whole thing.

The L3 switches will have a default route to the internet and static routes to point to the MPLS router as next-hop.

There is no protocol  running anywhere now .

Building 1 is the existing building of this customer and building two will be coming up soon .

What they want (save some cost as well) is :

   - Currently they have 2 MPLS at the existing site(Building1) which they want to split and move 1 link to the new site.

From there , the story I have just narrated follows.

They can not get the ISP to do any custom config so this is the issue they are faced with .

Once they split these links , how will these two sites be aware of each other and more over do failover.

Vlad

Hall of Fame Super Blue

Re: Redundancy between buildings

Vlad

Okay, if you are not exchanging routes with the ISP then you can do this. Lets assume site 1 is active and site 2 is passive in normal operations,.This also assumes the the 2 MPLS routers can route to each other via EIGRP.

1) Configure a static default-route on site 1 router that connects to MPLS pointing to the next-hop for the site 1 MPLS connection.

2) redistribute this static into EIGRP. Note this will mean the static has an AD of 170 on all other routers in both sites

3) Configure IP SLA on the site 1 router to track the availability of the MPLS link. You need this because you are not exchanging routes with your provider so you won't automatically know if the link has failed

4) On site 2 router that connects to MPLS  configure a floating static default-route pointing to the next-hop for the site 2 MPLS link. Note the AD of this route must be > 170

5) On the same site 2 router redistribute static into EIGRP.

This works as follows -

1) The site 1 router connecting to MPLS redistributes the static into EIGRP. All routers in both sites see this route and they will all see it with an AD of 170.

2) The site 2 router also sees this route and because it has a lower AD than it's own static route then it uses this one. It will not insert it's own static route into the routing table and hence into EIGRP because it's own static route has a higher AD.

3) Under normal conditions all traffic follows the default-route out via site 1.

4) If the link or the remote provider router fails then site 1 router removes the default-route from it's routing table. It no longer gets redistributed into EIGRP and so site 2 router now no longer receives it.

5) site 2 router now installs it's default-route and redistributes into EIGRP and all routers receive this route. Note all routers except site 2 will see this route with an AD of 170.

6) If the link comes back up at site 1 IP SLA will reinstall the removed static route and then this will get propogated to site 2 router. Site 2 router will then see it's own default route with an AD > 170 and the new one with an AD of 170 and will use the new one. So it removes it's own default route from the routing table and stops redistibuting into EIGRP.

7) All intermediate routers will now use site 1 again.

Note - when site 1 comes back up and installs it's own route and redistributes every intermediate router will have 2 default-routes in their routing table temporarily until the site 2 MPLS router receives the site 1 redistributed route. Once it does and it removes it's own default-route that will no longer be redistributed into EIGRP and so all the intermediate routers will only have one default-route left which will be via site 1.

And it would need testing but i have used this sort of setup before and it works fine.

Jon

Community Member

Re: Redundancy between buildings

Thank you very much Jon!

I will test this in GNS 3 and let you know!

Best Regards,

Vlad

365
Views
0
Helpful
8
Replies
CreatePlease to create content