[c-nsp] 6500/7600 split, Was Re: EOS/EOL for the 7500 platform

[Jared Mauch]

On Sat, Dec 16, 2006 at 12:25:11PM -0600, Dale W. Carder wrote:
> 
> On Dec 15, 2006, at 3:34 PM, Peter Salanki wrote:
> > I would say that the enterprise and service provider type of networks
> > are converging in technology
> <snip>
> > so I would rather expect a merge of the two BUs than
> > even more segregation.
> 
> Maybe I'm the only one who thinks the split might be in our best
> interest, because Cisco just can't pull it off.

	I think the chaos is going to quite possibly kill market
share.  Two different BU's "warring" with each other, neither
fitting into their market space perfectly.  I'm glad i'm not
a shareholder, it's a waste of money.  Any employee who
owns stock or anywhere else should be asking serious questions, IMO.

> The LAN feature set is huge.  The WAN feature set is huge.  So,
> in development of new software, the testing process is probably
> getting absolutely out of control.  I would guess that the "suits"
> realized that to be competitive, new features need to make it out
> the door faster.  To do this, split the software (and testing
> process) in half over time.
> 
> Look at the competition.  There's a vendor that makes a switch
> that can kick the 6500's butt.  There another vendor who's router
> kicks the 7600's butt.  If Cisco is forced to keep the "all in one"
> platform, the other two vendors will probably spank them because
> they can innovate faster within their speciality.

	There's a number of things that contributed to this "split".
There was a decision made by some managment that some infrasturcture needed
to change significantly for expansion.  A few other things happened
which caused the "7600" team to feel they couldn't get their
features in a shared release.  So they made the case of "lets do it
on our own".

> My guess is that the 6500 will continue to be *the* LAN switch/router
> platform, with just enough WAN features to get by.  The 7600 will try
> to be a Metro/WAN router with just enough LAN features to get by.

	Sure, except anyone that has seen the future knows that
stuff like fault tolerant software (ie: modular) is the way to go.
Look at the HFR, er CRS and what Juniper does.  As a SP I can't wait
5-7 minutes for my box to reboot because of some coding boo-boo.
Yes, software is imperfect, i know that, but there was this simple
technique called "protected memory" that came about.  It's not really
responsible to build a system these days without this fancy technology.

> It's great that the all-in-one platform has lasted this long.  I
> think it could have gone longer if CatOS were abandoned much sooner,
> if more effort were put into the PFC, and if the execs at Cisco could
> actually manage this big a team for this big a project.  Now, it's
> too late.

	As long as there is shared hardware, there will be the potential
for someone to make the case that this should be folded back in
together.  We'll see what happens in the next six months.  I have a
feeling some folks may get a surprise in the next month or so.

	- Jared

-- 
Jared Mauch  | pgp key available via finger from jared at puck.nether.net
clue++;      | http://puck.nether.net/~jared/  My statements are only mine.