DL to Core -- route or switch? 1

[rpast]

Hi folks –

I’ve been given a project to move vlans from our Core 6509s to the DL switches (3750s, etc. MLSs) in various closets, and have read two different ways of doing this. My question is which way is generally preferred.

The aspect I’m concerned with is the DL to Core connections. One way to connect is to configure point-to-point links between the DL and Core switches, with EIGRP or OSPF load-balancing between the cores. The other way is to configure a single vlan on the cores for all the various DL links going to the cores.

The second way seems odd to me, but I’ve seen it in at least a couple of publications.

Any thoughts on this?

Thank you.

 

[vipergg]

Actually it will work either way ,though I would prefer the second way it is less routed interfaces to deal with and might be a little faster though in this day and age most of the stuff is routed at wire speed anyway . The second way you just create a layer 3 svi on the 6509 and put all the conncting links to the 3700's as a layer 2 switchport . You then just do the same on the DL's. Create a new layer 3 SVI in that same vlan and put your uplinks on a layer 2 switchport in that vlan. The 3700's will all see each other as neighbors in this setup , basically what we use in a big corporate setup and it has worked well for like 7 years now .

 

[rpast]

Thanks for the response. So you’d have multiple SVIs in the same subnet in that same vlan? – I’m used to seeing one SVI per vlan. Couldn’t the L2 ports on the DLs just be associated with that vlan bounded on the Core, without defining an SVI on the DL?

Thanks again.

 

[vipergg]

You would have one SVI on the core in a certain vlan and then assign layer 2 switchports to feed your DL boxes and then put layer 3 svi's within that same subnet on each DL box.Thats the beauty of router/switches you can assign multiple ports on the core to the same vlan for the feeds to the DL . This would be your routed link from the DL to the core .

Example
Core

conf t
vlan 2


interface vlan 2
ip address 192.168.2.1 255.255.255.0

interface g1/1
descritpion feed to 3750#1
switchport
switchport access vlan 2

interface g1/2
description feed to 3750#2
switchport
switchport access vlan 2
--------------------------------------------------------

3750#1
conf t
vlan 2

interface vlan 2
ip address 192.168.2.3 255.255.255.0

interface g1/2
description feed to core G1/1
switchport
switchport access vlan 2

----------------------------------------------------------
3750#2

conf t
vlan 2

interface vlan 2
ip address 192.168.2.4 255.255.255.0

interface g1/1
description feed to core g1/2
switchport
switchport access vlan 2

 

[rpast]

So since I have redundant cores, I would create an extra L2 port on each DL for the second core, and keep hsrp running on the cores for that vlan?

 

[rpast]

PS - no way to load balance this way though, eh? except maybe through CEF load sharing?

 

[vipergg]

We actually use a second vlan in a different address range on the other side . If you do this it should give you 2 equal cost paths up to your core boxes . As you said CEF will load balance across these links . A good page to look at is here.

http://www.cisco.com/en/US/tech/tk827/tk831/technologies_tech_note09186a0080094806.shtml

 

[rpast]

Thanks for your suggestions. I've been seeing more scenarios w/ p2p routing, but this is something I will look into.

 

[rpast]

...one more thing.. can I assume that you do have HSRP running at both the DL and CL?

Thanks.

 

[vipergg]

On the uplinks there is no hsrp in our setup , they 2 individual routed links from the DL to the CORE. The redundancy is that you have 2 equal cost routed links , if one goes down it will take the other side.

 

[rpast]

thanks again.