[c-nsp] Cisco WS2960-X - Are these switches lemons or is there a stable release?

[Saku Ytti]

On (2015-10-12 09:34 -0700), Mike Hale wrote:

> I have a stack of 2 WS-C2960X-48FPD-L and they've been pretty solid.
> We're on c2960x-universalk9-mz.150-2.EX4.bin.

This is not targeted directly at you or the OP, but just an observation.

I believe stacks deliver inherently inferior service MTBF compared to
standalone switch.

Axioms:
a) each line of code have non-zero chance of having breaking bug
b) stack require more lines of code than standalone
c) software fails more often than hardware

So let's assume MTBF for the HW is 5 years, to remove that failure mode, you
install stack.
But in reality, if you consider MTBF of whole service, which includes software
defects, your MTBF is lower than 5 years already and you just reduced this
MTBF by increasing complexity by deploying stack.

Stacks are proprietary, as they don't need to interop, the amount of scrutiny
they receive is inherently less than interoperating features.

If you cannot deliver redundancy in service level, by client knowing how/when
to fall back to another server, then I'd consider introducing routing to the
service level, instead of stack/MCLAG etc.

-- 
  ++ytti