Routing with conditions

[trt]

Hi all,

I have a case that I had a few ideas in mind, but looking for the best and most efficient solution. Here's the case ;

We have 5 Cisco routers (1 central & 4 others connected to it) in 5 different locations which are connected via G.703. Each has 2 internal LAN interfaces (lets say NetA & NetB) and 1 serial. So we have 10 different internal LAN subnets, lets say ;

Region 1 - NetA :10.10.10.x/24              Region 1 - NetB :10.20.10.x/24       Serial : 192.168.10.1

Region 2 - NetA :10.10.20.x/24              Region 2 - NetB :10.20.20.x/24       Serial : 192.168.10.2

Region 3 - NetA :10.10.30.x/24              Region 3 - NetB :10.20.30.x/24       Serial : 192.168.10.3

Region 4 - NetA :10.10.40.x/24              Region 4 - NetB :10.20.40.x/24       Serial : 192.168.10.4

Region 5 - NetA :10.10.50.x/24              Region 5 - NetB :10.20.50.x/24       Serial : 192.168.10.5

Devices in 5 subnets of NetA will communicate within 5 regions, so do devices of NetB, but there will be no communication between NetA & NetB in both directions. Each client has two ethernet cards which are connected to the switches of NetA & NetB which was planned for redundancy, and these switches of NetA & NetB are not connected either.

So how can we apply dynamic route while announcing A & B networks from each router but limiting access from A-->B & B-->A by using minimum lines of access lists ?

Which routing protocol should be preferred, EIGRP or OSPF  ?

Thanks in advance.

Baris

[Vasilii Mikhailovskii]

Hello.

I would say that the design is really strange!

> Each client has two ethernet cards which are connected to the switches of NetA & NetB which was planned for redundancy.

If it's for redundancy, then what is one client lost NetA NIC, how should it communicate with all the other devices in NetA?

If it's for redundancy, then why don't you route between subnets?

If you want redundancy, then I would suggest to buy a couple of NICs that supports teaming. This would solve your "redundancy" problem.

Could you share - what is the background for such a requirements?

PS: why do you use single router per site (don't you need redundancy here)?

[Peter Koltl]

trt, do you have, by any chance, a background in storage/FC/SAN technology? :)
 

[Peter Koltl]

For complete separation I would consider

Easy Virtual Network—Simplifying Layer 3 Network Virtualization - Cisco

which requires Frame Relay or Ethernet VLAN WAN links.

[trt]

Hi Peter,

Does VRF run over serial links connected via VWIC2-MFT-G.703. The config is ;

controller e1 0/0/0

   channel-group 0 unframed

And serial IP addresses mentioned above are attached to the interfaces. I haven't applied clock source or timeslots configs. The G.703 port will be connected to radio link equipment.

Thanks

Baris

[trt]

Hi Peter,

Only FCoE but not Cisco. I have overall background in storage/FC/SAN, not deep technical.

Baris

[trt]

Hi Mikhailovsky,

Yes, the design is strange, that was my first reaction too.

If it's for redundancy, then what is one client lost NetA NIC, how should it communicate with all the other devices in NetA?

It will communicate through NetB, that's the point already, all devices will have connection to both networks using different subnets.

If it's for redundancy, then why don't you route between subnets?

Actually there's 1 device, connected to 2 networks carried by 1 router. Routing between two networks towards the same clients may result a confusing structure.

If you want redundancy, then I would suggest to buy a couple of NICs that supports teaming. This would solve your "redundancy" problem.

Server teaming configuration is not my responsibility, but it's not teamed right now. In both cases, with or without teaming the config should be running.

PS: why do you use single router per site (don't you need redundancy here)?

Router redundancy is two different ethernet NICs on each one, which the customer preferred.