Question regarding moving a virtual interface and VLAN's to a new device

Answered Question
Apr 21st, 2010

I am currently working to document out what it will take to move our corporate data center.  We currently have a 6509 as our server switch and will be uplinking connectivity from it to another location via dark fiber.  The original server switch will be going away once connectivity is established.

I will need to move the virtual interfaces over from the current production switch to the new switch, this would include modifying the routing as well I would believe.

The following is on the current switch:

interface Vlan15
description 550 Blair Mill Servers
ip address 10.125.15.3 255.255.255.0
glbp 1 ip 10.125.15.1

router ospf 1
router-id 10.125.2.11
log-adjacency-changes
passive-interface default
no passive-interface Vlan900
no passive-interface Port-channel2
no passive-interface Port-channel3
network 10.1.1.3 0.0.0.0 area 20
network 10.1.2.3 0.0.0.0 area 20
network 10.1.3.3 0.0.0.0 area 20
network 10.1.4.254 0.0.0.0 area 20
network 10.1.13.3 0.0.0.0 area 20
network 10.125.2.11 0.0.0.0 area 20
network 10.125.15.3 0.0.0.0 area 20

The current switch will have a fiber connection to the new switch which will need to house all the interfaces and the OSPF routing table.

I am just looking for some pointers on where to begin with this.  Any suggestions would be appreciated.

I have this problem too.
0 votes
Correct Answer by Jon Marshall about 6 years 7 months ago

Dennis

I think this is more a terminology issue.

SW1  <--- L2 trunk ---> SW2

In the above setup you can have a L3 SVI for vlan 10 on both switches. This vlan 10 could be a single /30 subnet ie.

SW1

int vlan 10

ip address 192.168.5.1 255.255.255.252

router ospf 10

network 192.168.5.0 0.0.0.3 area

SW2

int vlan 10 1

ip address 192.168.5.2 255.255.255.252

router ospf 10

network 192.168.5.0 0.0.0.3 area

the 2 switches will form an OSPF neigborship over the L2 trunk. And yes you can have passive-interfaces as well so you can passsive-interface all the other L3 SVIs bewteen the 2 switches so you don't get a large number of peerings.

Now say there is a vlan 20 on SW1 but not SW2

SW1 advertises out vlan 20

router ospf 10

network 192.168.6.0 0.0.0.255 area

so SW2 knows where to send traffic for vlan 20.  SW2 will see a next-hop of 192.168.5.1 on SW1. So SW2 receives a packet for vlan 20, routes it onto vlan 10 and then switches it across the L2 link. SW1 receives the packet on vlan 10 and then routes it to vlan 20.

If you have the same vlan in both sites it's even more straightforward because it only need to be routed onto that vlan on one switch and then it is all L2.

I think the confusion is because the idea of routing between sites suggests you need a L3 link.  What we are doing with a P2P vlan with a /30 is kind of emulating a routed link but because it is actually a L2 link you can still run vlans at L2 across that link as well.

Jon

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (1 ratings)
Loading.
Jon Marshall Wed, 04/21/2010 - 11:10

Dennis

Will the fiber connection be a L2 connection. If so you can simply run HSRP/GLBP between the 2 switches and then when the time comes to decommission the existing switch the clients will just use the new switch.


Only thing that's a bit confusing is your existing config ie.

interface Vlan15
description 550 Blair Mill Servers
ip address 10.125.15.3 255.255.255.0
glbp 1 ip 10.125.15.1

you have GLBP configured under the interface so are you running this switch with another for redundancy already ?

Jon

dennismatz Wed, 04/21/2010 - 11:15

Jon,

Yes, in our current data center we are running GLBP between two switches for redundancy.  We are decomming switch number tow before we begin this move so the GLBP will not be a factor.

I thought of using an L2 link and doing what you suggested but I have to configre the switch in the new data center to handle all of the OSPF routing, as the original switch will be gone.  We will also be working to move one VLAN at a time to decrease downtime.

Thanks,

-Dennis

Jon Marshall Wed, 04/21/2010 - 11:28

dennismatz wrote:

Jon,

Yes, in our current data center we are running GLBP between two switches for redundancy.  We are decomming switch number tow before we begin this move so the GLBP will not be a factor.

I thought of using an L2 link and doing what you suggested but I have to configre the switch in the new data center to handle all of the OSPF routing, as the original switch will be gone.  We will also be working to move one VLAN at a time to decrease downtime.

Thanks,

-Dennis

Dennis

You can easily have mutliple routers with GLBP so you could simply add the new switch. This would still allow you to move one vlan at a time.


As for the routing - can you provide a bit more detail as to why it wouldn't work ?

Jon

dennismatz Wed, 04/21/2010 - 11:50

Jon,

By no means was I saying it couldnt be done, basically all my thoughts keep taking me back to a L3 link between the two.  If I understand correctly you would suggest the following:

-Current redundant switch will be removed (as previously stated) making GLBP available to be used

-Create an L2 link between the two switches and configure as a trunk

-Create "router ospf 1" on the new switch in the new data center

-Create and configure virtual interface with required GLBP parameters

-Remove network statement from old OSPF routing configuration

-Add network to the OSPF routing configuration on new switch

Then continue those steps for each virtual interface/OSPF network

Not sure if that was what you were getting at, please let me know if I am incorrect.

Thanks,

-Dennis

Jon Marshall Wed, 04/21/2010 - 12:21

dennismatz wrote:

Jon,

By no means was I saying it couldnt be done, basically all my thoughts keep taking me back to a L3 link between the two.  If I understand correctly you would suggest the following:

-Current redundant switch will be removed (as previously stated) making GLBP available to be used

-Create an L2 link between the two switches and configure as a trunk

-Create "router ospf 1" on the new switch in the new data center

-Create and configure virtual interface with required GLBP parameters

-Remove network statement from old OSPF routing configuration

-Add network to the OSPF routing configuration on new switch

Then continue those steps for each virtual interface/OSPF network

Not sure if that was what you were getting at, please let me know if I am incorrect.

Thanks,

-Dennis

Dennis

Apologies for my wording, i was in a bit of a hurry. What i meant was could you just provide a few more details, i wasn't suggesting you were saying it couldn't be done

It all depends on what is connected to the new switch and when in terms of OSPF configuration. If you have external L3 devices currently receiving routes from the existing switches how will these be connected to the new switch ? Or is that not an issue.

If you have a L2 trunk you can always use that link to get to vlans on either side of the link ie. those that have been migrated and those that haven't.  So as you say, as you migrate each vlan, delete that network statement from the old switch and add to the new switch and while the migration is going on OSPF can simply route across the L2 link.

If you are sure you can migrate a whole vlan at a time then a L3 link between the switches would be fine. The advantage of a L2 link with GLBP/HSRP is primarily if you get caught out and need to have the same vlan in both places at the same time. It's kind of like a failsafe.

Jon

dennismatz Thu, 04/22/2010 - 07:14

Jon,

I have been testing the things we have been discussing.

I think doing a L2 link with the GLBP or HSRP would not be an issue.  My only hold up would be routing the OSPF over that link.  In the new data-center these 6509's will be handeling all the routing for the new building.  With that being said there will be other subnets behind the new switches that will have to torue back to the original data center/building.

With that being said above I think an L3 link would be required.  Correct me if I am wrong.  I would imagine that the OSPF network statements would have to be propogated once the VLAN (virtual interface) is moved.

Basically what is required is for the VLAN's to be available in both places, no matter where the virtual interface itself is.

Thanks Again for the assistance,

-Dennis

Jon Marshall Thu, 04/22/2010 - 10:21

dennismatz wrote:

Jon,

I have been testing the things we have been discussing.

I think doing a L2 link with the GLBP or HSRP would not be an issue.  My only hold up would be routing the OSPF over that link.  In the new data-center these 6509's will be handeling all the routing for the new building.  With that being said there will be other subnets behind the new switches that will have to torue back to the original data center/building.

With that being said above I think an L3 link would be required.  Correct me if I am wrong.  I would imagine that the OSPF network statements would have to be propogated once the VLAN (virtual interface) is moved.

Basically what is required is for the VLAN's to be available in both places, no matter where the virtual interface itself is.

Thanks Again for the assistance,

-Dennis

Dennis

If you want both vlans to be available in both places then L2 it is.

Still not sure why OSPF is a problem.

If you have a vlan in the old building that you need access to from the new building then simply advertise that vlan from the old building via OSPF and the new switches will learn of it and route traffic back to the old building.

If the vlan exists in both places but the device is in the old building then the new switches can advertise out the vlan and then switch any traffic for that device back across the link to the old building.

I get the feeling i'm not fully understanding you. If so could you perhaps give an example of why you think you need a L3 link between the switches ?

Jon

dennismatz Thu, 04/22/2010 - 10:34

Jon,

I was just not thinking an L2 link would work and was thinking L3 would be needed.  I was thinking this because I didnt think that OSPF would update the routes over the L2 link.

I tired to write out a brief config below, sometimes its easier to understand/determine with command syntax

Old Building Switch

==============

int g1/0/49

switchport trunk encapsulation dot1q

switchport mode trunk

router ospf 1

router-id 10.125.2.10

passive-interface default

no passive interface GigabitEthernet1/0/49 (can this be done if its not an L3 link?)

network 10.125.15.0 0.0.0.0 area 20

network 10.125.16.0 0.0.0.0 area 20

New Building Switch

===============

int g2/0/49

switchport trunk encapsulation dot1q

switchport  mode trunk

router ospf 1

router id 10.125.2.11

passive interface default

no passive interface GigabitEthernet2/0/49 (same thing, can this be done if its not  an L3 link?)

**remove network statements from old switch and poulate here when ready**

When the network statement is removed and then added to the new building I would imagine the virtual interfaces should be moved as well?

Let me know if that makes sense at all or am I entirely missing it.

-Dennis

Correct Answer
Jon Marshall Thu, 04/22/2010 - 11:30

Dennis

I think this is more a terminology issue.

SW1  <--- L2 trunk ---> SW2

In the above setup you can have a L3 SVI for vlan 10 on both switches. This vlan 10 could be a single /30 subnet ie.

SW1

int vlan 10

ip address 192.168.5.1 255.255.255.252

router ospf 10

network 192.168.5.0 0.0.0.3 area

SW2

int vlan 10 1

ip address 192.168.5.2 255.255.255.252

router ospf 10

network 192.168.5.0 0.0.0.3 area

the 2 switches will form an OSPF neigborship over the L2 trunk. And yes you can have passive-interfaces as well so you can passsive-interface all the other L3 SVIs bewteen the 2 switches so you don't get a large number of peerings.

Now say there is a vlan 20 on SW1 but not SW2

SW1 advertises out vlan 20

router ospf 10

network 192.168.6.0 0.0.0.255 area

so SW2 knows where to send traffic for vlan 20.  SW2 will see a next-hop of 192.168.5.1 on SW1. So SW2 receives a packet for vlan 20, routes it onto vlan 10 and then switches it across the L2 link. SW1 receives the packet on vlan 10 and then routes it to vlan 20.

If you have the same vlan in both sites it's even more straightforward because it only need to be routed onto that vlan on one switch and then it is all L2.

I think the confusion is because the idea of routing between sites suggests you need a L3 link.  What we are doing with a P2P vlan with a /30 is kind of emulating a routed link but because it is actually a L2 link you can still run vlans at L2 across that link as well.

Jon

Jon Marshall Thu, 04/22/2010 - 11:33

Dennis

Just a quick follow up. I've just noticed from your config that you are trying to passive-interface a switchport configured as a trunk which won't work.

Just in case i didn't explicitly state it, it isn't the physical port you peer over it is the L3 SVIs on each switch and that is how OSPF exchanges the routes so you would passive-interface L3 SVIs on each switch to stop them peering.

Jon

dennismatz Fri, 04/30/2010 - 06:01

Jon,

I apologize for not getting back to this right away.

I did as you suggested and everything worked great.  I created the /30 SVI and used it as the passive interface for OSPF and it routed without issue over L2 trunk link.  I cannot think of any other issue.

Thanks Again for all the assistance.

-Dennis

Actions

This Discussion

Related Content