[c-nsp] Opinions about the next 6500/7600

[Tony Varriale]

On 2/4/2011 4:27 PM, Mack McBride wrote:
> The cost per gigabit is not at parity yet for low gigabit rates.

If you are talking about 6500 vs N7K (which is what I thought we were 
discussing), then the N7K is cheaper.  And, so is the service.  Just 
simple math.

> The requirement for full IPv4 tables is governed by multi-homed bgp customers connected at the aggregation layer.

I wouldn't try to turn the N7K into an edge peering platform.

> As for the maturity level, we have a number of bug cases open on every platform we use.

Again, this goes for the 6500 as well.  I have a handful of bugs open on 
the N7K myself.  I don't notice the bug volume difference as long as you 
know what you are getting into.  I have stable N7K and 6500.  And I have 
wobbly too.

> We have customers that recently deployed N7K gear and were less than happy with the number of bugs encountered.

Yeah see above.  I suspect they are not working with a knowledgeable and 
experienced partner that is familiar with the ins and outs.

> We would rather not expose our customers to the less mature (and it is less mature) code.

In what manner?  Out in the street mature?  Sure.

Mature as in enterprise data center features?  Highly debatable.

> Curiously the ASR 1K which is a newer platform was very well conceived and has been relatively bug free.

I am a bit surprised but this is the general consensus I am hearing as well.

> Server refresh cycle is generally 18 months to 2 years while router refresh cycles are 5 to 10 years.

Generally true.

> And yes we have acquisitions with 10 year old hardware.

I noticed the key word there...acquisition.

tv