• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1318
  • Last Modified:

VLans and spanning tree

Hi,

I'm setting up a network based on L2 switches (Procurves 5304 xl).

We're using the VLans features of the procurves, by setting up on each one :
- a VLAN which will be dedicated to client conections
- 2 "routing" vlan for the connection to other switchs (Each Procurve will be connected to his 2 neighbours by Optical fiber, creating a loop.)

RIP is activated on each Vlan

My question is :
Do I have to activate spanning tree on the switches ?

I would say yes, because the loop topology is creating redondant connexions, but i'm not sure :
as there is only one direct connection between 2 vlans (even if there is 2 possible ways to reach it),  shouldn't we avoid the broadcast storm problem ?
Furthermore, vlans are stopping broadcasts...

Spanning tree is a problem for us : RIP is not really working well with STP. Because of the blocked ports, the best way is not always the one which will be chosen

Thanks for your help




0
Abgraal
Asked:
Abgraal
1 Solution
 
harbor235Commented:
If you have redundant paths, then yes you should enable spanning-tree, otherwise your network
will melt down. In addition, spanning-tree will allow you to efficently manage the flow of traffic
through your layer 2 infrastructure. Spanning-tree is a benefit in any layer 2 environment, I would never
consider a situation where I did not deploy it. In a situation where you did not deploy spanning-tree imagine
the scenario wher someone implements a layer2 device without yor knowledge, it could change the entire topology
of your layer2 network.

>Spanning tree is a problem for us : RIP is not really working well with STP. Because of the blocked ports, the best way is not always the one which will be chosen

You must make sure you have connectivity from any device in a particular vlan by setting up trunking dynamically
or statically. This ensures that you do not isolate a host on a particular vlan within your switch mesh. You can also
influence/change where blocking occurs by setting the root, and secondary bridges as well as manipulating
the path cost. Spanning-tree can be complex and most times is misunderstood. Hope this helps.


harbor235
0
 
pseudocyberCommented:
It depends.

We're a mostly Nortel shop and Nortel has a proprietary mechanism called Split Multi Link Trunking (SMLT) in which one switch can be connected to two switches in the middle and they in turn are connected to each other and there's no loop - a triangle - full mesh of 3.  With SMLT, the two switches appear to be one switch.

If HP has something similar then use that.  However, if they don't, then yes, you ought to use Spanning Tree at least.  You could get really creative and create seperate spanning tree groups and then have active/active redundant connections, but this can get really complicated - it's best to stick to the recommended design from your manufacturer.

Also, you can PROBABLY "bind" or ... Nortel calls it "trunk" cough cough ... interfaces together to form a faster connection.  Cisco calls it Fast Etherchannel.  So, you could have 2 GigE links which appear to be one link.  Then those two connect to another switch and provide you with a theoretical max connection of 2Gb - but the cool thing is if one fiber is disconnected, you haven't broken the Spanning Tree and everything keeps trucking.  So, you could have two bound channels of 2 GigE links each, right?  Then, if you're in a triangle/full mesh, one of those channels would be blocked somewhere depending on priority and MAC address.

In addition to running "plain jane STP", there's tweaks that are specific to manufacturer.  There's "rapid spanning tree", fast learning, etc.  Consult your documentation about these.

HTH
0
 
AbgraalAuthor Commented:
>You must make sure you have connectivity from any device in a particular vlan by setting up >trunking dynamically
>or statically. This ensures that you do not isolate a host on a particular vlan within your >switch mesh. You can also
>influence/change where blocking occurs by setting the root, and secondary bridges as well >as manipulating
>the path cost. Spanning-tree can be complex and most times is misunderstood. Hope this >helps.

Thanks for the answer, it helps.
And I agree, Spanning tree is complex...


Here's my config :

      -------------------- Procurve C
     |                                |
     |                                |
     |                                |
Procurve B                Procurve D ------- Procurve A
    |                                                           |
    -------------------------------------------------

RIP and Spanning tree are activated one the 4 servers
Procurve A is root spanning tree server.

The best link for Procurve C to reach procurve A is to cross Procurve B.
This is ok, but RSTP on procurve D block the port to Procurve C : then, procurve C have to cross Procurve B and Procurve A to reach Procurve C.

Is there a way to parameter RSTP then :
- Procurve C reach Proc.A using the Proc B link
- Procurve C reach Proc.D using the direct link

Thanks.

0

Featured Post

A Cyber Security RX to Protect Your Organization

Join us on December 13th for a webinar to learn how medical providers can defend against malware with a cyber security "Rx" that supports a healthy technology adoption plan for every healthcare organization.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now