LACP question

[jkuo]

We have a point to point T1 setup in bridge mode connecting two sites.  So there is no route, our LAN is basically extended to this other site for replication/DR.

Our replication data is completely saturating this T1 and we'd like to scale up the useable bandwidth by possibly adding another bridged T1, using LACP on 4200G switches at both sites to aggregate the two connections.

Is this even possible?  I'm not familiar enough with LACP or the T1 bridging to make a judgement.  Perhaps someone could help explain how LACP works, what layer does it work on?

Thanks!

My first post! Woohoo!

This message was edited by jkuo on 1-8-10 @ 12:09 PM

[Fred_Mancen_1]

Hi. The LACP feature works at Layer 2, so, teorically there is no problem to enable this feature in your environment. All you need to do is to create a link aggregation group, and assign it to the interfaces where the links are connected. I do prefer the static mode, so even when restarting the switch the group remains with the same ID.

In global mode:

link-aggregation group 1 mode static

link-aggregation group 1 description YOUR_DESCRIPTION

interface ethernet <interface_number>

port link-aggregation group 1

To check the feature:

display link-aggregation verbose 1

HTH

[jkuo]

Almost there.  The links seem to be in a failover mode where only one link or the other is being used at a time.

display link-aggregation verbose indicates that the Loadsharing Type is NonShared.

How would I implement loadsharing?

Thanks!

[Fred_Mancen_1]

You have switches on both sides, right? Enable the link-aggregation group on both switches. If the status is non load sharing this means that one of the ports is not active, connected or correctly configured. Make a double check in the configuration of the LAG, and implement the same feature in the peer switch.

HTH

[jkuo]

Using the same model 4200G's at both sites.  All port configurations look the same, but still one port is always unselected.  They are all configured as follows...

interface GigabitEthernet1/0/21


 stp edged-port enable


 port link-type trunk


 port trunk permit vlan 1 8 224 to 225


 broadcast-suppression pps 200


 undo jumboframe enable


 priority-trust cos


 lacp enable


 port link-aggregation group 1

interface GigabitEthernet1/0/22


 stp edged-port enable


 port link-type trunk


 port trunk permit vlan 1 8 224 to 225


 broadcast-suppression pps 200


 undo jumboframe enable


 priority-trust cos


 lacp enable


 port link-aggregation group 1

What other differences can I look for?  Thanks!

[Fred_Mancen_1]

Is there another LAGs created in on of the switches? The 4200G does not support too many groups (actually I think that only supports 4 groups).

At the other hand, if there are no more aggregations groups created, and since the settings are the same in both switches and in all the ports, I suggest you to check the physical connections as the transceivers, ports, patch cables and so on. Try to swap the cables in the switch in order to see if the same port will remains unselected.

HTH

[jkuo]

There were no other groups.  But decided to create a second group to another 4200G just to prove out that I'm creating the LAG correctly.  On that group, everything looks fine and I'm seeing both links as active.

Other difference.  In the LAG in question, it doesn't report the corresponding partner ports.  But my good group does report this.  I also don't get "neighbor" ports in the LAG in question, but my good group will report one.

Also, when swapping cables and ports.  The LAG in question seems to run the lower port number as selected and the higher port number as unselected.

Hopefully some info there will help?  Thanks!

[Fred_Mancen_1]

It seems that the peer switch does not have both ports correctly configured, so the LAG does not came up. At your side the settings are ok, but at the other side (the peer switch) it seems not to be ok.

HTH

[jkuo]

I control both switches, so I can verify that the switch port settings on both sides are identical.  Could the T1 "bridge" be dropping LACP?  I suppose I can contact the provider and tear apart bridge groups and try different things at that end.  Just want to make sure that we've exhausted options at the switch.

Thanks!

[Fred_Mancen_1]

Yes, it's possible...the LACP works at L2, so it's quite possible. If both switches are using the same settings and doesn't work, it's ok that you proceed a double check in the provider infrastructure.

Regards