Trunking/routing vlans from core switch puzzle

[wilson_1234_2]

We have two 6509 switches as our core network.

These two switches are a hsrp pair, one upstairs, one downstairs.

VTP configured as well, Upstairs switch is the server, downstairs switch is client.

The upstairs switch has a trunk configured to the downstairs 6509.

And a trunk configured to a 7206 router that connects our branches and dr site to the core switches.

OSPF is on the core switches and router.

We recently lost the upstairs 6509 switch.

In order to have the upstairs workstation/server links work while we were trying to get the switch replaced, we had to trunk two 3500 series switches to the downstairs 6509 and router in place of the upstairs 6509.

So, we ended up with:

One 3500 switch ended up as the other core switch, with a trunk to downstairs and a trunk to the router, with an additional trunk to another 3500 for additional ports.

There was no routing protocol on the 3500 series switches and we lost the VTP server.

The Downstairs 6509 was configured with all of the vlans and SVIs with inter vlan routing going on.

With no routing protocol on the 3500 switches (out of the box), what would have been the proper way to configure the 3500s to ensure that all vlans could communicate with the downstairs switch and branches?

The 3500s have servers and workstations. The workstations had to have DHCP addresses from the servers also.

workstations on vlan 10

servers on vlan 20

printers on vlan 30

routers on vlan 40

switch management on vlan 50

[wilson_1234_2]

Victor,

Thanks for your input.

I am trying to grasp the concept and I have an annoying charactersitic of my learning that I keep asking questions until I understand it.

So let me run down this again (your explanations are excellent by the way).

In my scenario in the 3550 switch that was trunked to the 6509, we are saying I only need layer 2 vlans and that if all ports are configured for workstations, that all workstations will "see" the SVI on the 6509 as their default gateway.

The same with the other switch that had Servers connected,which had a trunk from it to the first 3550 (trunked to the 6509), so it had to go through trunk #1 from itself to 3550 (trunked to the 6509) then through the second trunk, to the 6509 and it still would "see" the SVI of the server?

Is that correct? if so, I think I am getting it.

I could have put as many trunks in series that I wanted and as long as all vlans were trunked, I would get to he SVI on the 6509 from that respective vlan as long as the port was a memeber?

Correct and those are layer 2 vlans?

Please confirm the above.

If all of the above is correct, then I had a problem because I could not get an address from the server, even though ip helper was configured on the 6509 workstation vlan.

Also, I have an additional component to the scenario if you are up for it:

OSPF from the 6509.

[lamav]

Wilson:

I am trying to grasp the concept and I have an annoying charactersitic of my learning that I keep asking questions until I understand it.

That's OK. That's what you're supposed to do.

In my scenario in the 3550 switch that was trunked to the 6509, we are saying I only need layer 2 vlans and that if all ports are configured for workstations, that all workstations will "see" the SVI on the 6509 as their default gateway.

Yes, that's pretty much it.

You should understand that there isn't really a layer 2 vlan, per se, or a layer 3 vlan. A vlan is a vlan. It is a virtual LAN. There are different existential components of it, though. There is the layer 2/switched/ethernet component of it, and then there is the layer 3/routed/IP component.

So, depending on a switch's capabilities, you can, if you wanted to, configure the switch to perform both layer 2 and layer 3 functions of the vlan.

In your scenario, you have a 3550 and it has been decided by you that it will perform the layer 2 functions of the vlan. OK, no big deal. You create the vlan in layer 2, name it, and then place different access ports in that vlan. So, any workstation/server/IP host in general that is plugged into that switch port will be placed in the vlan for that port.

Now, since the switch is only performing the layer 2 component, the vlan's ethernet traffic will have to be sent "up" to the next layer switch to be routed so that users on that vlan can communicate with users on other vlans. On the switch performing the layer 3 functions of the vlan is where you will configure the layer 3 interface for that vlan.

The layer 2 interfaces were all the end user interfaces that devices were plugged into, as well as the trunk ports on the layer 2 switch.

The layer 3 interface on the switch that will perform the layer 3 functions of the vlan is called the SVI -- switched virtual interface. It is a layer 3/routed/IP interface with an IP address assigned to it. That interface will have a routing protocol running on it if you want to advertise reachability information for the network (vlan) it belongs to. I think that may be what you wanted to discuss regarding OSPF.

Think of the SVI as the gateway to the rest of the vlans. If traffic wants to leave or enter the vlan, it will pass through the SVI.

And you are correct again, the hosts on the vlan will "see" the SVI, meaning the SVI will be reachable through layer 2 because it's just another host on the vlan. An ethernet broadcast from one host will be "seen" by every other host on that vlan because a vlan is one broadcast domain. It's this ability to communicate over layer 2 with the SVI that allows it to obtain an IP address so that it can start communicating over layer 3.

I could have put as many trunks in series that I wanted and as long as all vlans were trunked, I would get to he SVI on the 6509 from that respective vlan as long as the port was a memeber?

Correct. You can extend the layer 2 domain with trunks over 3, 4, 5, 6 hops and more. So, a user on vlan 10 in switch 1 can communicate over layer 2 with a host in vlan 10 on switch 6.

If all of the above is correct, then I had a problem because I could not get an address from the server, even though ip helper was configured on the 6509 workstation vlan.

I do agree that you have a problem with users in a vlan being able to obtain an IP address, even though you do have the ip helper address configured. That is exactly why I joined in on this thread in the first place and why I asked you to post the configs of the 3500 switches, as well as the 6509.

HTH

Victor

[wilson_1234_2]

Thank you for the outstanding explanations,

I found out why the DHCP did not work:

The DHCP server was not configured on the downstairs SVI interface as ip-helper.

The only server that was configured, was not up when we were trying to get communication up with the 3550s.

I have a scenario concernig OSPF it you are up for it that happened that same night.

[lamav · Accepted solution]

Wilson:

I'm really glad you figured out the DHCP mystery! :-)

This was a really good VLAN thread. A lot of basic operational concepts were discussed, so I think you should mark the particular post that actually solved your problem, and Cisco will archive it.

That said, it may be a better idea to start a new thread for the OSPF scenario you want to present to everyone for analysis. This way the thread remains easy to follow along one specific track. Yes?

By the way, thank you kindly for your generous ratings.

Victor

[wilson_1234_2]

You bet Victor.

Thanks for the great answers.

[Jon Marshall]

Richard

Nothing wrong with asking questions :-)

Victor has given a very good explanation of how L2 vlans and L3 vlan interfaces work.

You are right in what you say, if you had 3 or 4 3550 switches all interconnected with trunks then assuming

1) the vlans exist at layer 2 on each switch ie. "sh vlan" shows the vlan existing

2) the vlans are allowed on all the trunks

then yes they would all be able to use the SVI on the 6500 as their default-gateway.

Vlans are layer 2 by definition. So just to clarify

"sh vlan" will show you the L2 vlans on a switch.

"sh ip int brief" would show you the L3 interface(s) on a switch.

As for OSPF bear in mind that it will only run on the 6500 and the WAN router as ll your 3550 switches will not be routing. So OSPF packets will simply use the 3550 switches to pass packets between the 6500 and the WAN router.

Jon

[wilson_1234_2]

Thanks jon,

Concerning OSPF, would the router and 6509 become OSPF neighbors and exchange LSAs with each other if they were several trunk links apart?

[Jon Marshall · Accepted solution]

As long as the vlan that the router and the 6500 have in common is allowed on all the trunk links then yes they will exchange LSA's.

Jon

[Istvan_Rabai]

Hi Richard,

Victor is telling you the right things.

Based on your previous post I assumed you already had the 3550 in layer2 mode, this is why I was confused about the behavior of your config.

You wrote the following:

"In the original scenario:

I have a 3500 trunked to a 6509 carying all vlans.

VTP configured and The 3500 is acting as a client."

Sorry for the inconvenience, and please do what Victor suggested you to do.

Thanks:

Istvan