up vote 0 down vote favorite
share [g+] share [fb]

Out of curiosity does anyone know whether it would be more beneficial to have more switches with less terminals on each leading to your main switch OR load up each switch fully before having any terminals added onto another switch?

link|improve this question
feedback

1 Answer

up vote 0 down vote accepted

Additional switches, whether they are full or not, will provide MORE paths to your upstream switch and thereby increase performance for users distributed across them. Many things will be a factor here as far as performance. Let's say your root/core switch is completely 1Gb, and to each downstream switch you provide a single link. Even though downstream each of those users is connected at 1Gb they are bottlenecked back to the core at the uplink. Now if you spread those connections out over multiple uplinks (read more switches) or by using link aggregation, you also increase performance. Really you need to define your requirements, but a bottom line answer is that "the more uplinks the better" regardless of how many switches you have or their port density.

link|improve this answer
1  
I'd also point out that massive uplinks and lots of switches won't help you much if the resource/server the user is communicating with cannot provide data that fast... so don't both building out a massive infrastructure if your users are accessing a NAS with two hard drives and a single 1Gb nic :) – SpacemanSpiff Nov 10 '11 at 16:56
=) Thanks, the main problem lies in the pipeline we have here in Florida and our main hub in North Carolina a measly 3MB connection :'( So I sought out to make here as efficient as possible. Interesting though with link aggregation which merely requires hooking up multiple connections between the switches and configuring them to identify these additional paths, I could increase the flow of traffic from areas switch after switch away from my core switch. With this you can do it in any configuration, correct? meaning 3 switches can connect core to a and b, then a connects to b? – 1Tguru4l1f3 Nov 10 '11 at 19:01
or should I say with switches core, a, b, c, d you can connect core to a and b, then connect c to a,b,d and d to a,b,c ensuring that more paths are created and if one link fails they can still make it to the core switch. – 1Tguru4l1f3 Nov 10 '11 at 19:05
well, you need to distinguish multiple paths (which spanning tree keeps under control to prevent loops) and link aggregation, which is combining multiple links from one switch to another. – SpacemanSpiff Nov 10 '11 at 21:39
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.