[arin-ppml] is NAT an inevitabile part of IPv4 / IPv6 transition
On 2/8/2011 8:46 PM, Lee Howard wrote:
> John, Tony, you are saying, "There is no way to avoid extensive deployment of
> large-scale NAT44 in ISP networks"?
> I have a hard time accepting that, since nobody wants it. It runs contrary to
> everyone's interest. It is a temporary solution at best, so companies have to
> deal with both LSN and IPv6, instead of just IPv6. Is everyone really resigned
> to this?
This isn't universal, but there will be significant amounts of NAT(of
various flavors) in ISP networks, especially telco networks. There are
as many issues with infrastructure gear as there are with customer side
equipment and largely for the exact same reasons (economics). This is
especially true of telco based networks since in many cases the
equipment has been in place for a decade or so and has been EoL'ed for
>=5 years. This _shouldn't_ be a problem but is because someone
involved in earlier DSLAM design decided that any IPv6 traffic must be
from bogons and decided to drop any frames with IPv6 (0x86DD) in the
EtherType field. Whoever first made the decision at this point doesn't
matter because that was copied by several different manufacturers so now
there a ton of DSLAMs (and I suspect early PON FTTx gear) that simply
won't pass layer 2 frames carrying IPv6 traffic unless its tunneled over
4. Whats worse because the gear is so old there isn't a
firmware/software fix available and in most cases simply won't be. This
doesn't include problems with DSL modems, most of which are routers,
which can't be upgraded remotely (if there is an upgrade) unless the
telco was very forward thinking and implemented TR-069. This also
doesn't include the fact that some of the most common lines of PPoE/oA
termination devices (Redback SMS line and AFAIK Nortel Shasta lines)
don't have an upgrade path. Redback (now owned by Ericsson) gleefully
points providers to their new line of gear (SmartEdge line) if they want
The equipment cost for one _small_ telco, ~3,000 DSL ports, can easily
exceed $1.75 million and that doesn't count the time and expense (and
customer disruption) to actually replace the gear. If they have to
replace modems on a large scale the cost will be at least triple that.
That means as the squeeze for IPv4 addresses starts to bite the cost for
doing CGNAT is far less than trying to actually fix the problem and the
vendors at least are claiming that most end users won't notice.
Vice President of Technology
ISP Alliance, Inc. DBA ZCorum
-------------- next part --------------
An HTML attachment was scrubbed...