[c-nsp] Cisco's new 4500-X 10G Aggregation Switches

[Gert Doering]

Hi,

On Fri, Feb 10, 2012 at 06:05:33PM -0500, harbor235 wrote:
> As far as the "tens of millions of routes" comment goes, my thoughts
> are along the lines of no real hardware out there designed for V6
> from the get go. Its all old V4 designed hardware retro-fitted for V6
> with a few exceptions.

No major surprise here, given that 99% of the traffic is IPv4 today, and
will be over 90% for the next few years.

[..]
> "Whoa, I seem to have missed something.  Last time I checked, we're
> at about 8000, and if we grow with the same speed (which is unlikely,
> but still) it will take about 30 more years to reach "tens of millions"..."
> 
> Outside the US there will be substantially more commitment to V6 and at a
> quicker pace, the PAC RIM comes to mind. 

Given the routing table numbers from the past 10 years, this doesn't hold
either :-)

See http://www.space.net/~gert/RIPE/weekly/ for daily-updated IPv6 
routing table views, split by RIR region and country inside the RIR
regions.  So far, the RIPE region is leading, with ARIN(!) following 
directly behind, and APNIC well below (~3000 routes/region vs. ~1000).

IPv6 deployment status is not directly reflected into the routing 
table - a region with 3 mega-ISPs might serve 2 billion customers and
push tbits of traffic, but show only up with 3 /19s in the global
routing table - while another region with many small ISPs might contribute 
500 /32s to the routing table, and only serve a few 10.000s of 
customers...  (which is, I think, what we're seeing now - many small
ISPs in RIPE land, fewer and bigger ones in APNIC).

> So when it goes it will not be at
> the
> same pace. I guess my complaint is there WILL BE incremental updates
> relatively soon. So........

Guessing from the IPv4 BGP table with about 40k active AS numbers
and an average of 1.6 routes per AS in the IPv6 BGP table, I expect
to see something like 50-75k routes in +5 years, and then a slowdown(!)
- of course my crystal ball is known to have been wrong regarding IPv6
predictions, but realistically, if we hit "10s of millions" of routes
in the global IPv6 table, something went seriously wrong on the way.

> This means I agree it will not have tens of millions V6 route capabilities,
> even at the edge.

There is no use in building a device today, with today's technology, for 
requirements that might *perhaps* show up in 20+ years.

gert
-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 305 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20120211/9c4357b1/attachment.sig>