I am pretty new to Cisco gear and am confused between the 3750 and 3560. It seems to me that the main difference is that the 3750s support stacking whereas the 3560s don't, but they do support clustering.
Can someone explain the difference between clustering / stacking, and why would I choose a 3750 over a 3560? Would the choice be simply down to stack-a-bility?
The 3750 stacking uses a high bandwidth link that produces a single virtual switch with a single configuration.
I forget the bandwidth but it is MUCH more that a few Gbits per sec.
3560 I guess (not 100% sure) uses GB Ethernet as the interlink.
If I was using 3560 I woud not bother with the clustering bit since the network performance will be the same without it and it is an additional complexity that is not required. I would consider using clustering if I needed to deploy a large number of sites each with a few switches. However I don't see much advantage - imagine dealing with a cluster member failure remotely with a non-expert technician on site pluging in the new switch.
With the 3750 it is integral to the package and works very nicely but you still would want to figure out in advance how to replace a stack member.
The issue I was trying to highlight is not related to some on-the-fly switch replacement during production but the process of replacing a failed switch. If I recall correctly new stack members automatically get a new stack member number however the central config refers to the old failed stack member so a new switch would need to be reassigned to the number of the failed switch.
A while back I ended up doing a TAC case when I wanted to add a 3750 to a single existing switch since I could not understand the documentation regarding stack member numbers and I did not want to risk the new switch becoming the "owner" of the configuration for the whole stack.
At the end of the day it works out OK but this was not spelled out in the docs.