Switch rack limit!!!!!!!!

[Hugo Tigre]

i've been searching for this but i didn't find much.

Is there a limit to the amount of cascade switchs.

I know the limit of hops, and i also know that i have to check the latency of the packet, depending on my topology.

but i'm not shure if there's some kind of limit to the number of switch i can interconnect!

thanks in advance for a reply

regards,
Hugo

[MIBA NSD]

Hi!
If u mean stacking I read that a Stack accepts up to 16 Switches!
regards peter

[Hugo Tigre]

Hi peter

i don't mean stacking, i just mean having the switches conneted to one another, for exemple:

host1 sends a packet to host2, then the packet has to travel through a number of switches to reach host2. I want to know if there's a limit to this number of switches, or if the limit is equal to the limit of hops (TTL).

regards hugo

[Caroline_7 · Accepted solution]

Hi Hugo

I think you can chain up to 8 switches or even more,in a star configuration, after that, you have to watch what you do, as you can have problems with speed, corruption and funtionality. If your network is large enough for 8 switches not to be enough for you, then you can turn one of your patch panels into a switch by connecting it to a port on one of the switches.

good luck
caroline

[Hugo Tigre]

Hi caroline

Thanks for the reply, thats exactly what i wanted to know, i can't really test that in my current lab environment, but i needed to have some idea for my report on the project i'm working on.

thanks for your help

regards,
Hugo

[Ardon]

Hi Hugo,

You have to make a distinction here between latency and the number of Hops.

You and Caroline mention a maximum of 8 which comes from running Spanning Tree and is actually 7. This means that two distinct bridges in the network should not be more
than seven hops away the one to the other.

Now another thing is latency. Looking at micro seconds (as switching is done in hardware ASICS) that will not impose any applicational problem. Normally higher Layer protocols (TCP or the Session Layer) can buffer Packets which arrive out of order and then assemble the complete one.

Believe me. I have seen networks with hundreds of Switches although a Frame will never travel trhough all of them of course but only a few and therefore latency can be ignored.

Regards, Ardon

[Ardon]

Forgot one thing.

You mention TTL. TTL is ONLY applicable when Routing comes into play (see my other posting "Number of hops can be reseted!!!!!!!!!! " )

At Layer 2 (Switching) regular Frames carrying user traffic have no Layer 3 header and therefore no TTL.

[Hugo Tigre]

Hi Ardon,

thanks for the reply, your answer really help me.

I do know the difference between latency and hops! What i didn't know is what's responsible for restricting the number of cascade switches.

One thing u said i didn't quite understand:

- "You and Caroline mention a maximum of 8 which comes from running Spanning Tree and is actually 7.This means that two distinct bridges in the network should not be more
than seven hops away the one to the other."

Basicly what i think u r saying is that Spanning Tree is the one responsably for allowing a packet to only run through a total of 7 Swicthes!!!! But, Why is Spanning Tree responsible for this?

regards, hugo.

[Anders_35]

The 7 hop limit is an IEEE recommendation, not an absolute.
It has to do with the time-out of root information (max age) in the spanning-tree.
See http://www.cisco.com/warp/public/473/122.html for more information.

[Les Ligetfalvy]

When it comes to meshing, HP has a 5 hop rule.

[Sergej Gurenko]

Use RSTP or MSTP and you will newer face theoretical "7 switches limit" problem.

[Hugo Tigre]

Thank you all for your input

it really helped.