Catalyst 3524, 3550 Failover

[us2001irvine]

I have one Cisco Catalyst 3524 XL core switch (8 fiber gigabit port) connecting the company's network.

I was asked to add another switch to provide link redundancy. That is, in case the core switch (CAT 3524) fails, the other switch will automatically detect and bring up the line.

Right now I have a Catalyst 3550 switch: 10 RJ45 gigabit ports and 2 fiber gigabit ports.

Do you think that it is possible? And how to configure it?

I really appreciate your help. Thanks.

 

[baddos]

It is possible only if your fiber runs are close enough to run your copper ports w/o media converters. It would be best to keep your links standardized though.

 

[us2001irvine]

Thank you, Baddos. Actually, I can stack the two switches together on the server rack. May I ask how to connect and configure this two switch for failover? Thanks.

 

[baddos]

It depends on how your network is configured...

Also, remember that the 3508 is a layer2 switch and the 3550 is a layer 3 switch. The 3508 will never be able to do routing when the 3550 goes down.

 

[lui3]

problem here is with the redundancy in the fiber links. if you have two links from each closet running into the core then you could terminate them into two switches. with two 3524s or a 3550 with fiber ports you could do this.

the setup would look like the following
xxxxxxxxxxxxxxxxxxxxxxxx
3524 x uplink1 uplink 2 x 3550
xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxx
x x
x x
x x
x x
x x
x x
x x
x x
x x
x x
x x
fiber link 1 x x fiber link 2
x x
x x
x x
x x
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx

closet switch stack

You would setup the 3550 as the secondary root switch and make the 3524 primary root

you could also set the port cost of link 2 from the closet switch stack to 500. this would force the port to block for spanning tree purposes.

no need to connect the 3524 to the 3550 in the core. keep the 3550 configured as layer 2 with no ip routing command configured.

no you have a failover situation handled via spanning tree. when one link fails the other will recover within 45 seconds. if you use uplink fast on the ports in addition to spanning tree you will get less than 5 second failover.

this is one way to do it. and is a common cisco design for redundancy. point is you need more fiber links from the closets for failover.

 

[us2001irvine]

Thank you Baddos and Lui3.

I need to re-define the problem statement as I get more info for the project: I have a network spreading on two floors in a building. Each floor has a Cisco Catalyst 3508 8-port fiber core switch connecting each sub-switch. The problem or risk for this design is that if the core switch is down then that floor network will be down and if the core switch in the server room which is connecting the other floor is down, then both the two floors network will be down. So, I would like to provide link redudancy at core switch level.

Right now I run Cisco 3508 8-port fiber switch as core switch. I would like to have 3550 10-port fiber switch or higher for the failover. Can anyone give some comments on this in terms of hardware requirements and setup configurations. (I need to use fiber to connect the core switch in the server room to the other floor for both main and failover core switches. There is one thing that I am not sure is that how to connect the two core switches on the same floor. Do we use crossover cable to hook it up like what I do for Cisco PIX 525?)



Fiber connect
Floor_1 (server room) ----> Floor_2 (Core SW)
Floor_1 (server room) ----> Floor_2 (Failover SW)

SW1(failover)--SW1(main) ------> SW2(main)--SW2(failover)
| Cross connect? | | Cross connect? |
| | | |
| | | |
sub_sw sub_sw sub_sw sub-sw



Thanks a lot

 

[baddos]

Probably hook up your switches in a configuration similar to this if I am reading your post correctly.

SW1 - SW2 (First Floor)
| |
| |
SW3 - SW4 (Second Floor)

The PIX 525 can support failover if you purchase the Failover box for about $4000 and your have the unrestricted OS license. If you have another PIX 525 for failover, then plug one PIX into one switch and the failover into the other.

 

[us2001irvine]

Thank you Baddos.

I think that you might be right. However I am not quite sure. In this failover senario, there are two points of failover: Core switch level and the two floor fiber connection level.

For Core Switch Level: we may connect the two core switch with fiber port while to stackwise the two core switch may not be right because the two switches will be seen as one switch.

For Floor Fiber Connection: I have already had one pair of fiber to connection the main core switches between the two floors. I am supposed that I need to connect failover core switch with another pair of fiber between the two floors. This is logical but I am not sure how the failover technology will make it work for the physical setup in this senario.

Thank you.

 

[baddos]

If you look at my diagram, the switches are connected in a loop. In this configuration there is no SINGLE point of failure, but it's not as fault tolerant as a full mesh topoligy.

 

[us2001irvine]

Yes, Baddos. This is what I am thinking too. I hope that my vendor can help me out for the implementation and I will get back to this list as I put this project through.

Thanks once again, and
Have a great day.