what is difference between Switch stacking, etherchannel? Can anyone e

[jobsp90]

If there are 2 switches stacked uplinks given to each 2 switches with 2 ports each, say eg port 47 & 48.Port 47 & 48 are created LACP for each switch Then what is the use of stacking here for redundancy?I am newbie. Please explain.

[MHM Cisco World]

Without stack the other peer see four neighbor ID for one lacp link

But with stack the other peer see one neighbor ID and hence lacp is work

MHM

[Joseph W. Doherty]

If your example you're not obtaining the optimal redundancy of the stack and LACP combination configuration.

Consider if you, instead, configured both member switches ports 47 in one LACP group and ports 48 into the other LACP group.  Do you see the advantage?

[M02@rt37]

Hello @jobsp90 

In a network setup where you have two switches stacked together and each switch is connected to two separate uplinks using ports 47 and 48, which are configured for LACP, stacking and LACP work together to provide both redundancy and increased bandwidth.

When two switches are stacked, they act as a single logical switch. Stacking provides several advantages, particularly in terms of simplified management and hardware redundancy. You only need to configure and manage the stack as a single device, even though it consists of two physical switches. If one of the switches in the stack fails, the other switch can continue to operate, ensuring the network remains functional without any significant disruption. Stacking also enables you to extend the switching capacity of the network without adding complexity, as both switches work together seamlessly.

In your case, ports 47 and 48 are used in an LACP port channel (also known as an etherchannel) for each switch, which allows multiple physical links to be aggregated into a single logical link. LACP helps increase bandwidth by allowing the two physical connections to work as one, distributing traffic across the links. Additionally, LACP provides link redundancy: if one link in the aggregation fails, traffic automatically switches to the other link without any disruption, maintaining network availability.

Now, when you combine stacking with LACP, you achieve multi-layer redundancy. With stacking, the two switches operate as a single unit, and if one of the switches fails, the other switch continues to function. At the same time, the LACP configuration on ports 47 and 48 ensures that if one of the physical uplinks fails, the traffic can still use the other uplink. This setup enhances both switch-level redundancy (thanks to stacking...) and link-level redundancy (thanks to LACP...).

The key advantage of stacking here is that it allows for a more resilient setup. Without stacking, if a switch were to fail, the entire set of connections to that switch would go down. But with stacking, even if one switch fails, the surviving switch can take over the workload, keeping the network running without needing to manually switch connections. The LACP link aggregation further ensures that traffic is spread across multiple links for both redundancy and improved performance.

[13jobsp90]

No my doubt is if I remove stacking, but both the switches are having separate uplinks from core switch so if 1 switch goes down, the other will work because it also has got an uplink to the core swtich.So here what is the use of stacking other than acting as whole single unit?

[Joseph W. Doherty]

What happens in one of your core switches fail?  I.e. to the independent switch that just lost its Etherchannel uplink.  Wouldn't all the connections in that switch lose network access (to the rest of the network)?

Of course that independent switch could have other infrastructure links, but that's one of the features of a stack.

In many ways, a stack is much like a chassis with multiple line cards, multiple sups/DFCs, multiple power supplies and increased bandwidth between stack members.  Also, as one logical device, again like a chassis, it reduces L2 nodes, L3 nodes (if L3 capable), device node management.

Basically, stacks offer many of the advantages of using a chassis.

Even for large chassis platforms, so many of these advantages go hand-in-hand with larger port count supporting devices, such as 6513 chassis supporting 11 96 PoE port line cards, 6500s supporting IAs with up to 2,016 host ports, Nexus supporting FEXs (low thousands?) or ACI (several thousand?).

[MHM Cisco World]

it seem you dont want to bay money for stack 
so you can use below instead of stack 

[13jobsp90]

yes.this is what i need. Stacking is not required other than as a single entity.

[MHM Cisco World]

you can use above 
PO between standalone Acc SW and Core SW give you 
1- high BW 
2- high availability (if one link of PO is down) 

this design not need Stack and there is no issue with STP if there is only one Core SW and there is no interconnect link between Acc SW 

MHM

[13jobsp90]

no there are 2 core switches and 2 access switches in my scenario and servers which are connected to access switches.

[MHM Cisco World]

this for two Acc SW and two Core SW 
note:- there is no interconnect between Cores 
note:- there is one uplink is BLK by STP 

check this if you have Q ask 

MHM

[Joseph W. Doherty]

BTW, the example L2 topology design provided by @MHM Cisco World is functional but usually considered poor.

(Also BTW, as diagram actually shows edge devices are L3 switches, you could do edge routing which often would be considered the best approach, if it could be used.)

Why a poor L2 design?  Traffic between cores needs to transit an access edge switch.  Traffic between one core and one access switch may need to transit the other core and another access switch.  An example of the latter would be communication between Core1 and AccSW2.

The "fix", to avoid using an edge switch for transit, is to provide an additional link between the cores.

With the core<>core link, you can avoid using an edge switch for transit but you can also avoid using a core switch for transit too.  The latter requires every individual VLAN is only known to the two cores and only just one edge switch.

If just one edge switch per VLAN, you make it the STP root, otherwise the gateway hosting switch should be the STP root.

The one VLAN per edge, if using GLBP, should provide the best (L2) load distribution.