L3 Redundancy vs Stacking

[hs08]

Hello,

Which one is better for redundancy L3 between use Stacking or use HSRP/GLBP ?

[Reza Sharifi]

Hi,

Stacking usually is a better choice, because you are managing one logical switch using one management IP.

HTH

[Richard Burts]

The original post asks about " redundancy L3". How does stacking provide any redundancy?

From a management perspective I can see that a stack that treats multiple switches as a single unit might be desirable. But to provide L3 redundancy clearly HSRP/GLBP is the better option.

[MHM Cisco World]

I think your Q must be correct to stackwise virtual vs HSRP 

for me I prefer HSRP but Cisco recommend stackwise virtual, why ?
there is some tech that use the stackwise virtual  benefit like SSO NFS Q-SUP

[Joseph W. Doherty]

Could you clarify your question as HSRP/GLBP is generally not used for L3 redundancy?

Stacking can be used instead of multiple switches and/or multiple routers.  Pros and cons to both, including on the subject of redundancy.

[fabioaquino13255]

HS08,

Before answering, I find it interesting to detail a small difference between stacking functionality and other means of redundancy.

Stacking will actually give you a higher density of ports, as the junction of 2 or more switches will operate as a single one. This makes it easy to manage.
Let's assume that you have 24 points in your network and this size is left over, so you don't have to manage 2 switches, since you are the only network admin, this will generate less work and will meet your demand, since you will have 2 switches of 24 ports operating as 1 single port of 48. This stacking can be undone at any time. However if a switch like that comes to a problem you lose 24 connections. This doesn't give you redundancy, to have redundancy you would have to have 2 more switches stacked in the same way.

High availability is the ability to ensure the continuity of used services, even in times of failure.
Redundancy is linked to high availability implementations. Basically it refers to the presence of additional devices to be adopted as a backup or balancer.
So if we look at the above definition of stacking, we cannot consider it as a redundancy solution.

Hope this helps.

[Joseph W. Doherty]

@fabioaquino13255 do I understand correctly, you're saying a switch stack offers no additional redundancy, it only provides easier management?

If so, by the same logic, a chassis with multiple sups and power supplies (in redundancy mode) also provides no additional redundancy?

[fabioaquino13255]

In my understanding yes. Why? Let's assume you stacked a new 24-port switch in your topology, now having 48 ports to be able to add 6 new equipment, since the old switch had no more ports, if this new switch burns out 1 week later, where will it be? the redundancy? The equipment did not flow the traffic through the other.
If I'm wrong, please correct me.

[Joseph W. Doherty]

Poor redundancy example.  Why I write that, how does your example differ from adding a second stand-alone switch, to connect six new connections which fails after a week?  More or less no major difference.  Understand, regarding redundancy, there's much overlap between chassis platforms with redundant hardware, independent stand alone switches and stacked switches (also VSS and vPC).  Each has its pro and cons, and each often supports some redundancy to preclude single points of failures dropping network access for one or more devices.

Perhaps a better example, you need 20 ports, but being so close to 24, you decide to obtain a 48 port switch to allow for port expansion.  You do that, and that switch fails, you lose all 20 connections.

So, you decide to obtain two 24 port switches, placing 10 connections on each.  One switch fails, 10 connections stay up and while waiting to repair/replace failed switch, you can repatch to the running switch.  Notice, I didn't compare whether the two 24 port switches are stacked or not because just considering loss of a live port, alone, they're about equal.

Also, to keep devices connecting to the dual switches, we might keep them on-line if we can dual connect them, i.e. a connection to both switches.  Again, might be done with switches stacked or not, although with switches stacked, you might be able to take advantage of Etherchannel.

For basic redundany, again, you avoid single points of failure.  However, stacks do offer other advantages over stand alone switches, such as mentioned in other replies, one (logical) device to manage, or in this reply, Etherchannel, but there are other pros too.  Again, stacked switches offer many pros, but they have their own cons.

To recap, stacked switches can provide some redundancy, but so can multiple stand alone switches.  Actually, some redundancy of a switch stack is provided, implicitly, by a switch stack having multiple stack members.  A stand alone individual switch, though, generally can offer much the same, regarding redundancy, when you have multiple stand alone switches in parallel (not series).

[fabioaquino13255]

I understood your explanation Joseph.
I put my answer based on the options our friend HS asked, that's why I put it that way, we learn together from it.
Thanks.

[hs08]

Hello,

I believe with stacking we also get L3 redundancy, example we have 2 cores switch running using stacking then we must have 2 uplink for each Distribution Switch and connect to each cores. With that topology i belive if one core failed, the another core still running normally alsfo for the DS. Am i right?

[Joseph W. Doherty]

Stacking L3 switches vs. being stand alone are much the same pros and cons when only L2, especially with regard to redundancy.

However other issues and/or same issues but weighted differently influence making such a choice.