[c-nsp] Solid L2 switch - 2948G or 3548-XL-EN?

[Neal R]

  I'll second the negative report on the 2948G-L3. I've ran into one
(actually a 2980G-L3) in the field and it managed to create a routing
loop using two interfaces *that were both running OSPF*.  No, I'm not
kidding, I got access to another one later and could reproduce the
problem - it was a total delight to troubleshoot.

   The last time I was in Optimum Data there were dozens of those things
stacked up ... the word is they're an L2 switch or they're bought as a
maintenance part to replace one that has failed. Nobody turns 'em up for
routing if they can at all avoid it.

    Buy a 3550, get something later than 12.2.25SEE so rapid spanning
tree is actually rapid, and you'll be much happier.




Ian Dickinson wrote:
> Lamar Owen wrote:
>   
>> On Saturday 23 June 2007, sthaug at nethelp.no wrote:
>>     
>>> Plenty of bugs. *Especially* if you actually tried to use them for L2.
>>> The 2948G-L3 is dead, and deservedly so.
>>>       
>> Ok, what sort of bugs have people experienced with these?  I say this because 
>> I have three, and if there are configurations that tickle these bugs, I'd 
>> like to know so that I can avoid those configurations.  Not using them is not 
>> an option, though.
>>     
>
> Anything that used ACLs was suspect.  The L2 support was very weak.  We were
> so impressed with the 2948G-L3, we took them out of service, put them in store
> until they'd depreciated, and then put them in the bin.  I'm not sure I can
> think of a configuration that isn't inherently flawed in some way.  It's by
> far and away the worst product I've ever used that had the Cisco logo on it
> (though some unlucky souls may say the NSE-1 was worse, I never had one).
>