I have been working
to configure a stack of 3 WS-C3750X-48T-L so 2 members are active and 1 is in hot standby
mode. From the best I can understand reading and rereading the manuals standby
switches that would roll in to take the place of a failed switch isn't possible
with stacks.
It appears to make
this work you must use switch clustering HSRP groups. I am confused about
clustering and cluster command switches. Can one of the active 3750's act as a
command switch. If so then can a cluster of 3 be configured so 2 are active and
upon a failure the standby takes the configuration of the failed switch?
Essentially you have
two choices based on what you're looking
for. For the automatic configuration of a replacement stack member, you would
actually need to leave your stacks at 2 switches. The automatic reconfiguration
comes when a switch fails, and you physically remove and replace it with
another switch. The new switch will detect that another switch is already
running in the master role and will accept the existing configuration.
Your other choice is
to have a stack of 3, with one of the switches running empty. Upon a failure,
use the current running configuration to copy, replace interface numbering, and
paste in to place. Then move the cables.
There is no way to
have a stack member automatically take on the config of a failed member... At
least none that I know of. The better answer to this is to get your customers
to do, well, what they currently aren't doing which is dual connecting their
equipment.
(Caveat to my
opinions - I've never done clustering myself, we've always used either stacks,
6500's in pairs, redundant SUPs, or other means of providing HA capabilities)
I don't believe that
clustering would solve your problem from my reading of Cisco's docs... The two WS-C3750X-24P-S switch cluster that rauenpc
suggests looks like your best bet. If
you're really worried about swap out time, you could rack three switches in and
simply leave one disconnected. The
amount of time it takes to replug the network cables and stack cables shouldn't
be that significant especially if you leave the middle of the three as the cold
standby unit that way you'll only ever remove the top or the bottom switch
rather than pulling out the middle switch (if your top switch dies and the
bottom of the three is the standby, your re-cabling gets messy real
quickly).
Honestly, I have to
agree with rauenpc's recommendation on driving your customers to "do the
right thing". And, unless your SLAs
are written to require this kind of failover, you're probably costing yourself
more money trying to compensate for a customer trying to do it on the cheap.
没有评论:
发表评论