migrate network segment to vlan routing with dlsw+

[ccsam]

Currently, the dlsw are terminated on the core router which has token ring interface. The rest of the network are segmented using ethernet interface on the core router.

I need to migrate the network segment on the core router to cat6509 which will run inter-vlan routing.

My concern is when I move down the bridge group from the router ethernet, any issue will arise?

core router core router

| |

cat6509--------cat6509

[tblancha]

Is it going to be the same vlan/ethernet segment that will be bridged into DLSW? But you want the bridge-group and the dlsw to terminate into a different router? If so, you should remove the bridge-group from where it is first and then add it to the location where you want it second. Then give CAM a couple of minutes to switchover and everything will be OK.

The bridge-group command needs to be on the router terminating DLSW and be on the vlan or ethernet interface that has layer2 access to the SNA devices.

You might be running IEEE spanning-tree in the router which might not be an option on a native IOS on the 6509-use vlan-bridge instead.

[ccsam]

Here is my setup topology :

R1---Host

/ \

CS1 CS2

\ /

DS1

I have DLSW peer from CS1 and CS2 ( 6509) to R1. My client connected on the DS1 vlan 200. There is a etherchannel connection between CS1 and CS2. Everything works fine, the client can connect to host and I can see the circuit.

CS1 is the root for vlan 200. When the link between the CS1 and DS1 failed, the client cannot establish circuit to the host. I cannot see the MAC address of the host, it is possible due to dlsw redundancy issue?

I did remove the dlsw peering from CS2 but it still not working.

[mbinzer]

Hi,

sorry your topology is not clear to me. Can you clearify a bit more what the current situation is and what you are intending to do.

On what lan media are the end systems currently connected, including the host.

What redundancy issue are you refering to? Something on the etherchannel or do you have two dlsw peers up and running on the same ethernet vlan?

thanks...

Matthias

[ccsam]

Hi,

My HQ router running SRB/TB to allow ethernet segment at HQ router to access IBM host. There is an active dlsw peer to BR1 and backup peer to BR2.

L1 and L2 are links between gigabit interface to routed port at core switch which running bridge group 1.

CS2 will be the root for vlan 20,40,200. CS1 will be the root for vlan 10,30,100.

L3 and L4 are trunk link to CS1 and CS2. AS1 is connected to DS1 on a port assigned to VLAN200. SNA client is on vlan 200.

Vlan 20,40,200 are assigned to bridge group 1.

For normal operation, SNA client establish circuit to the host. From CS1, I can see host mac address learn through bridge which is from gigabit interface connected to BR1. From CS2, I can see the host mac address learn through port channel.

When L4 link down, the circuit still maintain with interruption. This time CS2 does not display any host mac address due DS1 will forward vlan 200 to CS1. The SNA traffic will go direct to BR1.

During L4 recover ( meaning reconnect back the lost connection), I can see host mac address learn through vlan 20,40,200.

After the mac aging, all mac address disappear and I can’t establish any SNA session. What could be the problem for this?

[jishi]

You really should use DLSw Ethernet Redundancy in your case.

Here is an example:

http://www.cisco.com/en/US/tech/tk331/tk336/technologies_configuration_example09186a0080093ecc.shtml

DLSw ER should fix your problem.

Jing

[ccsam]

Hi,

Basically, the dlsw peering will primary will always to active and backup will alway passive without connected until the primary dlsw peer failed.

Ethernet redundancy require router to map host address to M' and M'' for host Mac address M. If I have lots of client and server need to connect to host, then I need to change all my configuration on end station. Any other way to do this.

[jishi]

You can use real address map, i.e. M' to M', if you don't care about the loadsharing, then you don't need to change anything on your clients. Since in your current setup, the primary is always active and backup is always passive until the primary fails, you are not loadsharing anyway.

If you don't want to use DLSw ER, you have to add filters to prevent loop and the mac address flipping between switch ports.

[ccsam]

I would like to know what do you mean by mac address flipping?

When refering to filter, does it mean MAC address filtering and input/output lsap filter? I did try using output-lsap-list the result quite consistent where my client (connected at access switch) at vlan 200 and CS2 will learn back the host mac address through vlan 200.

When running without lsap filter, the MAC address sometimes learn through vlan 20 and 40. After that, all connection will lost. I do not know how to explain the behaviour on this?

Basically, vlan 20 will be in forwarding state on bridging spanning tree at both core switch. When VLAN 200 spanning tree changes, why does it affect the bridging spanning tree?

regards,

[jvogler]

Remember some basic legacy rules

1) only one entry path into the DLSW network without DLSW redundancy. The fact you are blackholing everything indicates you are breaking this rule. The remote traffic MAC address gets cached as local by the switch from explorer traffic from the host loops back out of the network. Remember how basic switching works. The source mac is cached.

2) DLSW does not flip the mac address. It sends everything in non-canonical format. You can eliminate the translational bridging router by peering your 6500 to the Token ring router. This will facilitate communication to the TR segments without translational bridging.

Here is a nice link on how to enable DSLW at the 6500

http://www.cisco.com/en/US/partner/tech/tk331/tk336/technologies_q_and_a_item09186a00800949ff.shtml

The link has a nice sample for a 6500/MSFC to a router. This should allow you to manipulate the configuration to get your situation functioning.