Switches, Hubs, and Modems
Showing results for 
Search instead for 
Did you mean: 

Buildning redundant core with 3500 as stacking

Buildning redundant core with 3500 as stacking


I want to build a redundant core with 3500 or 5400.
Is it possible to build a redundant core if i stack 2 3500-48ports, and setup a mesh-ring to each floor.
I need to have about 15 VLANs routed in 3500.

The other solutions is to setup VRRP and use 3500 or 5400, but I think it´s easier to just stack both swtiches, and I don´t need the premium license.

Totally I will have about 11 HP 2626, 6 HP 2810-48ports for Edge, and 5 HP 2900-24 ports for Servers.

Any suggestions ?

Best Regards, Magnus Tengmo
Trusted Contributor

Re: Buildning redundant core with 3500 as stacking

Layer 2 meshing is quicker than Layer 3 VRRP, since the devices do not have to openup the frame, read Layer 3 info, make routing decissionc and then wrap the Layer 3 packets back in Layer 2 frames before pushing these on the Layer 1 media (wires or wireless).

So I'd go with the mesh-ringing that you mentioned.
Mohieddin Kharnoub
Honored Contributor

Re: Buildning redundant core with 3500 as stacking


Meshing and Routing can't be together, and since you have 15 Vlans, then you need routing between them, hence you can't use meshing.

Using the 5400 better, since its modular and you can integrate many solutions in this core.

Using the 3500 is not a bad idea, but if you calculated the cost of a 3500-48G and a 5400 Bundle like 5406-48G, the bundle wins.

Adding L2 redudancy can accelerate you time of recovery using Spanning Tree.

L3 redudancy is required also on the core if you are planning for dual cores, and you can use the VRRP after adding the premium edge license.

Also, you can add the VRRP redudancy later on if you are not interested in L3 redudancy, and just depend on the MSTP to load balance the whole network on both Cores.

Good Luck !!!
Science for Everyone

Re: Buildning redundant core with 3500 as stacking

Thanks for the input. Probably 3500-24 is enough for us if using meshing.
Is it possible to have two 3500-24 in core setup with meshing for L2 redundancy, and these two connected to 6400s for L3 redudancy with 10gb ?
6400 got XRRP, any advantage with XRRP over VRRP ?

If we will go with the 5400s, probably the spanning-tree loop will have at most 8 switches, isn´t that too much?
Will the spanning-tree loop be smaller if we stack the edge switches to one stack / floor ?

Best Regards, Magnus