Does a stack switch provide fault tolerance, advice on Stack Vs Spanning Tree for HA

[Kdes]

Good Day All

By no means a network expert so thought ill ask here .

We moving offices and gives us an opportunity to redo our LAN to a much better standard. The switches we buying are x15 Aruba 2930F 48G PoE+ 4SFP+ Swch for about 400 nodes.

My thinking is to have two switches stacked as a core switch, then have the other 13 switches connected to each switch in the 2 switch stack.

SW1 - SW2 (Stack)
Core

SW3 - SW13
Access
Each connected to SW1 and SW2 ie one cable to SW1 and one cable to SW2

My questions are;
From my understanding a stacked switch acts as one switch, when one of the pairs fail. Will the other continue operating as normal?

Can i have a aggregate port from SW1 and SW2?
eg SW1 Port 1 and SW2 Port 2 aggregated to connect to a access switch?

My thinking is to use aggregated links from the access switches to the core stacked switch to;
a) provide fault tolerance if one of the core switches fail.
b) will give 2Gbps throughput.

Didn't want to use spanning tree because it seems to be more for stopping loops rather than meant to be used for High Availability unless i am wrong and is the correct way to go?

Thank you.

 

[syntax]

Stacking essentially takes the management plane and spreads it across both switches. For all intents and purposes, it functions as a single switch and should be thought as such.
Some considerations are how does the switch stack (special stacking cables with high throughput or normal interfaces).
How do the switches get firmware upgraded, can they do in service upgrades to avoid downtime etc

You would normally aggregate access switch to a port each on the stack. You are incorrect on the 2Gbps of throughput, you have 2Gbps of bandwidth, but throughput would still be 1Gbps.

STP stops loops whilst creating redundancy. You do however usually waste a link because it gets blocked until another looped link fails.

 

[EasyUp Web Hosting]

In a Cisco stack, the other switch(es) will continue to function, if one of them fail, even if it was the first switch in the stack. Not sure about your switches though. Also, are you sure those switches are stackable? I find any info about that on those switches.

 

[irBosOtter]

In a Cisco stack, the other switch(es) will continue to function, if one of them fail, even if it was the first switch in the stack. Not sure about your switches though. Also, are you sure those switches are stackable? I find any info about that on those switches.

They support VFS
Virtual Switching Framework, so you stack them that way

@OP
Are all the switches the same model? You don't need layer 3 switches for access switches, unless you want to use L3 on them I suppose.
Maybe buy cheaper layer 2 switches?

 

[Kdes]

Thank you.
Yip they all the same model some details below.
https://www.hpe.com/us/en/product-c...pip.aruba-2930f-switch-series.1008995333.html

What is the typical way that switches are setup? Are they normally just all connected to one switch like a star topology?

 

[irBosOtter]

Thank you.
Yip they all the same model some details below.
https://www.hpe.com/us/en/product-c...pip.aruba-2930f-switch-series.1008995333.html

What is the typical way that switches are setup? Are they normally just all connected to one switch like a star topology?

Suppose it depends on budget

But try to follow the "Core" "Distribution" and "Access" layer model.

But a lot of smaller companies skip the distribution stack if the environment is small.
Try not piggyback switches off each other. All access switches should be plugged into the dist stack, or core stack if no dist stack at least

 

[syntax]

Thank you.
Yip they all the same model some details below.
https://www.hpe.com/us/en/product-c...pip.aruba-2930f-switch-series.1008995333.html

What is the typical way that switches are setup? Are they normally just all connected to one switch like a star topology?

As said, standard is 3 tier or 2 tier model. These now reflect physical more than logical, as layer 3 often terminates on the firewall for security purposes.

Your access switch is usually dual connected, either to a stack, VSS type switch or 2 separate switches with spanning-tree blocking one of the ports.

 

[TheGuy]

Is using vrrp not a beter idea than stacking switches?

We ran vrrp on our core switches and then the access switches had 2 ports in a port channel connected to each switch. Our VMware environment was also split across both switches. This meant we could take a core switch down without causing any downtime.

 

[syntax]

Is using vrrp not a beter idea than stacking switches?

We ran vrrp on our core switches and then the access switches had 2 ports in a port channel connected to each switch. Our VMware environment was also split across both switches. This meant we could take a core switch down without causing any downtime.

if you port channel from access to 2 separate core switches, you either need them stacked or running something like VPC.
VRRP is adding redundancy at layer 3, stacking / VPC is doing this via layer 2. It would all depend on your design and where the layer 3 terminates, from your scenario it sounds like you have a collapsed core, although the port-channel to 2 switches makes me think it must be running some form of VPC.