spanning tree query

[knaik99]

1.in real life,

why is it good to avoid STP and go for ethechannel?

I want answer wrt loop. is it difficult to trace loop if there is loop in STP ?

If we have site with 2 cores and 35 access switches and if we find loop in such lan network,is it difficult to trace ?

because of that we have to avoid STP

[balaji.bandi]

Layer 2 domain hard to find the issue and there no easy to fix if anything go wrong, so that is the reason STP is evil always. ( you can configured span gauerd and loop prevention).

But i would suggest best solution is   Core to Access to have Layer 3 P2P link with OSPF or EIGRP is best approach(make access switch as layer 2 where the end device connected.)

Or do you have any reason all need to Layer 2 domain.

[knaik99]

means you are telling me to run OSPF between access switch and core

what if 10 access switches are using same Vlan 10 with subnet 10.x.x.x/24 and every access switch is going to cores (2 core--both are at different location)

in such scenario we can not go for routing between access switches and core as I have to keep SVI of vlan 10 on core only 

[balaji.bandi]

In this case your Layer 2 Stretched across Access switch, (i was in impression you doing new setup)

There are 2 Options.

Option1 :

If you like to run Core and Access with OSPF, you need to re-do IP address scheme, this is for 1 time and Later you have more piece of mind.

Since these are only Internal IP you can change with DHCP reservation, if they are reserved IP and you want to retain same IP (then you need to deploy Layyer 2

Option2 :

Layer2 between Core and Access switches, Core will have you Layer 3 Interface you can Port-channel between Core and Access switch

You can use additional STP protect mechanism :

https://www.ciscopress.com/articles/article.asp?p=2995351&seqNum=3

why is it good to avoid STP and go for ethechannel?

if you deploy 2 Trunk Link STP detect Loop and 1 of the port get in to Block mode state as the convergence take place.

if you build Port-channel you have advantage of use both the ports  and higher throughput and resilient link available.

is this make sense ?

[MHM Cisco World]

Ether channel make you use all link while the STP will block one link to prevent the Loop.

detect loop is simple but solve loop is issue because you need to shut down the link and solve the loop.

[Mohsin Alam]

STP loops are not that hard to trace , you can always track down TCN changes using on CAT switches :

show spanning-tree detail | inc ieee|occurr|from|is exec

Also you can run ether channel for redundancy or you can use STP variants to have different switches as root for different VLANS.

If you follow all best practice for STP you should not run into issue , e.g access port with port fast and BPDU guard .

 
## Make sure to mark post as helpful, If it resolved your issue. ##

[Joseph W. Doherty]

The two main advantages of Etherchannel over STP are: Etherchannel failover is faster, and, as already mentioned by others, you may get to use all the multiple links bandwidth (instead of "redundant" links being blocked - although with Cisco's PVSTP, you can assign different links to different VLANs).

The difficulty of "tracing" L2 loops with STP, if you have one, STP has "failed" and your network might go into L2 meltdown.

Even if you want "avoid" STP, it's good practice to have it active in case some creates an accidental L2 loop.

Also, when using STP, recommend a "rapid" variant (not the default on Cisco switch [or at least it wasn't]).

[Jon Marshall]

Just to clarify, it is not an either STP or etherchannel choice (unless you are running L3 etherchannels only). 

Even with etherchannel you should always run STP in a network with redundant links and even one without. 

Jon