Unsolved

This post is more than 5 years old

184 Posts

12254

January 29th, 2005 14:00

Spanning tree woes

Hello everyone, i have run into a problem with the spanning protocol on some dell 3048's. We have 48 users connected to the switch and the 2 gigE uplinks trunked to a 24 port gigE root switch(which is partially managed). If an effort to prevent a specific group of users from saturating the gigE links we decided to vlan ports 1-24 to port 49(gige) and 25-48 to port 50(gige). So we have vlan 1 and 2 now. When the switches are powered up the spanning tree protocol blocks one of the gige connections. My understanding of the spanning tree protocol is that it wants to prevent loops in the network. However this is technically not a loop. When i disable spanning tree everything works as it should. computer on port1 cant talk to port 48 without going through the gigE links. If i am misunderstanding something out the spanning tree protocol can someone please explain it to me. I dont really like to run without the spanning tree enabled.

What i think is happening is the switch bridge can see itself through both gigE links and see's that as a loop, when in reality it is split into 2 vlans. Thanks in advance for any responses.

2 Intern

 • 

812 Posts

January 31st, 2005 12:00

Currently, none of the PowerConnect products support a Per-VLAN Spanning Tree. All PowerConnect switches support Spanning Tree based on the IEEE 802.1d (some models also support 802.1w) standard. This provides a single Spanning Tree per the physical bridge.
 
If your uplinked switch supports IEEE 802.1q frame tagging, you could create a Link Aggregation group with the 2 Gigabit ports and trunk both VLANs across the Link Aggregation group. This would not dedicate bandwidth for each VLAN, but it would increase your bandwidth and provide redundancy in case one of the uplinks fail.

184 Posts

February 1st, 2005 16:00

Thanks for the reply. That basically answered my question. What i think may happen is spanning tree will be disabled on the edge switches 3024's/3048's. But the core switch is going to change to a powerconnect 5324/5224. I am hoping we can leave the spanning tree protocol on for this core switch so we can at least use the broadcast storm control functionality. Or is the core switch spanning tree going to see the same problem with the 2 gige links from one switch?

2 Intern

 • 

812 Posts

February 2nd, 2005 10:00

Spanning Tree and Broadcast Storm Control are two different features and do not provide the same function. The 5324 and 5224 also only support a single Spanning Tree per bridge. If you connect more then one port between these switches (without configuring Link Aggregation), Spanning Tree will block the redundant link.

If you disable Spanning Tree and rely on the Broadcast Storm Control, you will cause the loss of legitimate traffic. Broadcast Storm Control does not prevent a storm from occurring. It simply drops all broadcasts over the specified threshold to prevent the storm from crashing the switch.

As in the previous scenario, your best bet would be to use Link Aggregation between the switches to increase the available bandwidth and 802.1Q frame tagging to pass traffic from both VLANs across the uplink.

184 Posts

February 2nd, 2005 12:00

Thank you for your help gregh. I guess i was misunderstanding how broadcast storm control worked. I thought it was a function of the spanning tree. In that case we will probably give up on the vlan idea, since it wont give us the poor mans bandwidth control we are after. we will just trunk the gigE ports, with 1 vlan per switch allowing us to keep spanning tree and broadcast storm control turned on. Thank you once again.

0 events found

No Events found!

Top