Same OSPF area 0, EIGRP between them

[Fernando Hernández]

Hello.

My network is fully EIGRP at the Data Center, but there is going to be another brand integrated at the DC, so need to use OSPF on different blocks.

I would like to understand what could be the problems I could face if I use the next topology:

[OSPF Área 0]  Block 1 ←→ (redistribution) ←→ [EIGRP - Switches Core] ←→ (redistribution) ←→ [OSPF Área 0] Block 2


or 

[OSPF Área 1]  Block 1 ←→ (redistribution) ←→ [EIGRP - Switches Core] ←→ (redistribution) ←→ [OSPF Área 2] Block 2

I understand that I may encounter routing loops if appropriate filters are not used, but something else I should take into consideration?

[paul driver]

Hello
Nothing they will be seen as two different networks  due to eigrp in between them 

[MHM Cisco World]

[MHM Cisco World]

case 1 using same ospf process and area 0 you need as I mentioned GRE 
case 2 using differnet ospf process and any area you can safe redistribute from and to eigrp , lab above for you 

MHM

[Joseph W. Doherty]

---

@MHM Cisco World wrote:

case 2 using differnet ospf process and any area you can safe redistribute from and to eigrp , lab above for you 

---

As long as the two (or three) EIGRP<>OSPF redistributions are on different physical routers, OSPF process ID can also be the same (as it's local to just that router).

For example, using:

[OSPF Área 0]  Block 1 ←→ (redistribution) ←→ [EIGRP - Switches Core] ←→ (redistribution) ←→ [OSPF Área 0] Block 2

Physically Brand X <> Cisco <> Brand X

As Brand X, if it doesn't support EIGRP, the single Cisco device would need to do the redistributions, and for that, you need distinct OSPF process ID numbers.

If physically Brand X <> Cisco <> Cisco <> Brand X

Then each Cisco device can use the same OSPF process ID number.

For either of the above, again, OSPF area numbers don't matter.

BTW, for:

---

@MHM Cisco World wrote:

case 1 using same ospf process and area 0 you need as I mentioned GRE 

---

You don't necessarily need to use area zero on a GRE tunnel as long as the GRE area and block areas all use the same area number.

However, as noted by OP, as the OSPF routers need to fully (routing) integrate with the EIGRP network, I don't see any major advantage of using a tunnel, but I do see disadvantages.  That noted, a tunnel could be used.

[MHM Cisco World]

@Fernando Hernández @Joseph W. Doherty 

There are three cases 

1- case 1 

Using same ospf process area 0' 

Again you need GRE to solve split area 0 of same ospf process

2- case 2 (optimal design)

Different ospf process different or same area 

I show in lab it work there is no problem 

3- case 3 

Same ospf process different area' can not answer that 

What point make design of case1 and case3 issue 

A- multi redistrubte point which cause loop or non optimal path 

B- ospf is interconnect via other link

C- split area 0 

So we need more info from @Fernando Hernández 

He is sure there is no other link interconnect ospf 

What is ospf process he use 

MHM

[Joseph W. Doherty]

As I understand OP information, there's no specific need to directly join the multiple OSPF blocks into a single OSPF domain.  The need, I believe, is just being able for traffic to correctly route throughout the whole network, which will contain both EIGRP and OSPF routing protocols.

Many of your responses focus on using a common OSPF process ID.  Again, the OSPF process ID is local to the router, it doesn't matter to OSPF neighbor routers in the same OSPF domain.  It does matter to different OSPF domains on the same router.  (One way to think of multiple OSPF process IDs, on the same router, it's like having different routing protocols on the same router.)

Likewise, you focus on using area zero.  It's only needed to share OSPF routes between different OSPF areas.

For instance, the OSPF blocks would work fine, via a tunnel, as long as the all had the same area number, which wouldn't need to be area zero.

As mentioned in an earlier reply I see little to no advantage of using a tunnel.

Although OSPF process IDs and area numbering is likely inconsequential, routing between EIGRP and OSPF is not.

Again, if the OSPF are edge networks, likely all they need is default route to the EIGRP network.  The EIGRP network will need to know the OSPF networks.

[MHM Cisco World]

Let postponed reply to each other until we get more detail from @Fernando Hernández 

For GRE' i write three points a'b'c about points of design.

Thanks 

Sauron 

[Richard Burts]

MHM seems to believe that there is a requirement that all OSPF routers see all of the OSPF routes as internal OSPF routes. And he proposes solutions to achieve this. I do not see it that way (and I believe that Joseph does not see that). I looked at the OP and focus on this part of it "there is going to be another brand integrated at the DC". I take that to mean that there are going to be some devices introduced into the network that do not support EIGRP (non Cisco devices) and which will run OSPF.  

I do not see any requirement that OSPF devices communicate with other OSPF devices using Internal OSPF. And I note that while the tunneling OSPF solution does provide direct communication between OSPF devices, it does not provide any communication with the EIGRP part of the network and the OSPF part of the network. To achieve communication between OSPF and EIGRP there needs to be redistribution. And if they are going to need to redistribute between OSPF and EIGRP then what is the point of introducing OSPF tunnels?

I do agree that it would be good if @Fernando Hernández would provide some clarification. 

[Joseph W. Doherty]

Oh, either a point I haven't made yet, or I didn't make significant, if you do redistribution, between EIGRP and OSPF, it will need to be done on a Cisco device, assuming Brand X doesn't support EIGRP.

All the redistribution could be done on just the Cisco device or, EIGRP<>OSPF redistribution still on the Cisco device in conjunction with additional redistribution, OSPF<>OSPF on Brand X.