cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1890
Views
0
Helpful
7
Replies

L3 switching & STP

a.pijpaert
Level 1
Level 1

Hi,

i'm pretty new to this spanning tree protocol and layer 3 switching but I was wondering...

If i have a redundant hierarchical network - with layer 3 switches in core/distribution and access layer, would I still need the STP or RSTP protocol to prevent loops? Or is this resolved by the routing protocols?

Thanx in advance...

7 Replies 7

ywadhavk
Cisco Employee
Cisco Employee

Hi Alvin,

You should not disable the STP. Please refer to the below url for more details on the importance of a STP.

http://www.cisco.com/en/US/partner/tech/tk389/tk621/technologies_configuration_example09186a008009467c.shtml

STP works at the switch port level unlike the routing protocol.

Hope this helps a bit.

Thanks,

yatin

Hello Yatin,

thank you for your fast response to my question, unfortunatley I cannot acces the page your link refers to. I've contacted cisco for access to the page but I've been told that I need partner or distributor level access. Unfortunatley I'm a student working on my final paper and this information could help me a lot. Is there a way I can still access this information?

Hope you can help me,

thanks in advance,

Alvin

Alvin,

Try this:

Understanding and Configuring Spanning Tree Protocol (STP) on Catalyst Switches

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_configuration_example09186a008009467c.shtml

Same URL as posted previously, except the word "partner" is removed.

Hope this helps.

Hi yatin, or anyone else...

What exactly do you mean that "STP works at switch port level unlike the routing protocol"?

I understand the way STP works but I'm not sure why we can't 'get rid of' STP.

Correct me if I'm wrong but this is the way that I see it:

If a packet enters a L3 switch, the switch looks up its destination in the routing table and sends the packet over the correct interface. And since there is only one entry in the table, the packet is send over the correct interface to it's destination. So no loops.... right?

Also layer 3 switches, like routers, decrement the TTL field in the IP header so once that drops to zero the packet is dropped anyways right? So no packets will travel on the network forever.

And in the case of using the STP, there will always be 'lost' bandwidth on the blocked link right? Is there any other way of using the redundant link without the use of VLAN's?

Do I make a point here or are my thoughts leading me in the wrong direction?

Many thanks in advance,

Alvin.

I believe that the reference to "STP working at the port level unlike the routing protocol" is that, if you use your L3 switches as L2 switches as well as L3 routers, you can run into an instance where a physical link may go down (which STP would correct for) but your VLAN stays up, so HSRP failover never takes place.

I had a very detailed discussion with another gentleman in here not too far back, talking about how you CAN have L3 redundancy without L2 STP protection. The bottom line was, if you want to avoid STP, then you have to treat your L3 switches as L3 routers only; that is, you can't have multiple physical ports on one L3 switch belonging to the same VLAN.

As for lost bandwidth on the blocked STP link, if you have multiple VLANs you can engineer it so that some links are blocked for certain VLANs but not others, and vice versa. Cisco calls it Spanning Tree Load Balancing, and it's a byproduct of being able to do Spanning Trees on a VLAN by VLAN basis. Of course, this works best in an all-Cisco environment.

If you're mixing equipment from different vendors, that complicates things. Your best bet then for sanity is to get by with a blockage on one link that is the same blocked link for all VLANs. Many vendors only support one instance of Spanning Tree per switch, which is the limiting factor here.

If you are interested in learning everything you ever wanted to know about STP but were afraid to ask, an excellent reference is Cisco Press's "Cisco LAN Switching." If you go to http://www.ciscopress.com and find that book, the last time I checked BOTH chapters on Spanning Tree were available as on-line samples. Worthwhile reading...I used to avoid STP, thought it was simplistic and outdated compared to L3 redundancy. I have come to realize that with new enhancements, it does things that are still needed in today's networks, and takes care of situations that L3 redundancy doesn't address. It's a lot of work to really understand your networks and get them tuned, but it's worth it in the long run.

Hope this helps.

From my perspective I wouldn't disable STP unless I was the ONLY one to ever work on a network (and of course I would never leave the company or be sick :-) ). STP is looking at mac addresses. If someone mistakenly connects another switch or hub into a network and then creates a loop then your LAN will fall apart quickly and resolving may take some digging. Also, without STP you cannot setup redundant links to a router or another switch without causing a loop (of course assuming you want redundancy). Hope this helps.

Actually, you CAN set up redundant links between a router and a switch , or between two switches, without causing a loop if STP is disabled, if those devices can do Cisco EtherChannel or IEEE 802.3ad link aggregation. These technologies create a single logical connection out of multiple physical connections; the limitation is, they must be direct connections between those two devices.

If you have redundant connections and some of them are indirect, then you need STP. For example: three switches in a "triangle" configuration, where each switch connects to two others; for each switch ("corner" of the triangle), there is a link that is not directly connected to it (the "side" opposite the "corner") that nevertheless is providing a redundant pathway for the network.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: