HELP! how to configure a network so that if one switch link falls the second one will be used

[NordWest1]

Hi All, 

I am a beginner in networking and at the moment I am facing some problems with my network configuration.

I have attached the network design. Considering the design I have chosen for my purpose how can I configure the router and Core switched (multilayer switches) so that if one falls for some reasons the second one is up and takes over. 

Please, help me. I have already set on both switches server vtp mode and spanning tree different so that the first switch will have priority 0 the second one 4096. 

I will try to elaborate more on the problem I am facing. 

Initially, my network design was looking like this 

I also want to add that it is configured fully and everything works as required. I asked my tutor for feedback and he said that it would look better if I could replace the core switch with a multilayer switch. I did so and I configured my multilayer switch everything works. Now he said I should I should add another one just to increase my redundancy. And here is the point where I got lost. How should I connect the second switch at the core level so that my redundancy will increase?

1. should I connect the second switch to the router if yes (how should I configure the router or the port I have connected the second switch)?

2. should I connect the second switch just to the first switch and set priority 4096?

[Gary Archer]

Is this for a theoretical exam question or would you like a real world answer from the field ? 

[NordWest1]

Yes, that's for a theoretical exam, thanks.

[Gary Archer]

It looks like you have a crazy amount of links in this. There are a couple of clues in the diagram that can point you on the best route to take.  Spanning tree is a layer 2 solution and is quite limited in what it can do in this scenario dur to the number of links you have.

This needs a layer 3 solution with a dynamic routing protocol to deal with any failover.

The clues in the question are:

• Each department has been given a separate subnet so L3 routing will be needed to reach them
• The 2960-X is a layer 2 / 3 switch which supports OSPF
• More than two paths to a switch through multiple layers makes an L2 solution impracticable

The best solution is to use OSPF layer 3 routing at the core and distribution layer and layer 2 at the access layer

1. Remember with OSPF “All areas must connect to area 0”
2. Place OSPF Area 0 between core and distribution layers. This will allow multiple other areas for your external router and access switches
3. Try to Configure all links as point to point /30 IP addresses, this will avoid layer 2 looping problems and make things a little easier to manage.
4. Use “no switchport” on the switch interfaces to put them layer 3 routing mode
5. Try to group your area link addresses together this will make summarizing multiple links in the same area easier in your configuration

Good Luck with your certification journey

[Gary Archer]

You probably wouldn't do it like this in the real world.  For link resiliency to a switch you can use etherchannel, you don't have to run the separate etherchannel links on the same internal cables routes for extra protection.

If this is combined with crossstack or card on chassis based switches in the core or distribution layers you can guard against switch failure at those layers.

Or just use SD access with a layer 3 fabric layer and layer 2 overlay. but that's a whole other discussion.

[Dennis Mink]

I suggest you verify STP and bring the primary link down in your lab to see if STP recalculation occurs.

[Georg Pauwen]

As stated by others already, spanning tree will take care of the redundancy, as long as everything is cross connnected to each other. With regard to router redundancy, since you only have one router, the only way I can see to create redundancy is through an EEM script that tracks the connected interface, and then if that interface fails, recreates the same interface configuration on the backup interface.