----- Original Message -----
From: "Jeroen Massar" <[EMAIL PROTECTED]>
To: "'Brian McGehee'" <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Wednesday, April 02, 2003 4:14 AM
Subject: RE: NAT != SL (Was: RE: CONSENSUS CALL: Deprecating Site-Local
Addressing)


> Brian McGehee [mailto:[EMAIL PROTECTED] wrote:
>
> > *Notes below
> >
> > * I agree "NAT != SL"!   Except in the case that hosts need "global"
> > connectivity and they ONLY have  a SL address will require
> > NAT.  Or they should have a second IPv6 globally unique unicast
> address
> > (which isn't that hard to have multiple addresses???)
>
> Use a firewall to block incoming packets.
>
> > * I can envision an enterprise environment that is comfortable using
> > FEC0::/10 for internal communications but a host has a globally unique
> > address also for external communications (on hosts that have this
> > neccessity)  This could be comfortable to a network admin/ops/noc.
>
> Then don't route a certain prefix.
Which prefix?  SL Prefix?  Where is the alternative?
>
> > ----- Original Message -----
> > From: "Jeroen Massar" <[EMAIL PROTECTED]>
> > To: "'Brian McGehee'" <[EMAIL PROTECTED]>;
> > <[EMAIL PROTECTED]>
> > Sent: Wednesday, April 02, 2003 2:53 AM
> > Subject: NAT != SL (Was: RE: CONSENSUS CALL: Deprecating Site-Local
> > Addressing)
> >
> >
> > > Brian McGehee write:
> > >
> > > > NO -- Do not deprecate site-local unicast addressing
> > > >
> > > > - Site-locals should be retained for disconnected sites.  If
> > > > not SL, then
> > > > a mechanism needs to be adopted that can provide a
> > private means of
> > > > selecting from a private address space that is "reserved" for this
> > > > function.  2002 is not a working alternative.
> > > > - Site-locals should be retained for intermittently
> > connected sites.
> > > > - Site-locals should be retained as a means for internal
> > > > connections to
> > > > survive global prefix renumbering.
> > >
> > > All of the above will make the address not unique.
> > > Have fun merging your networks.
> >
> > * Anyone that has had to merge 2 very large networks will
> > know the great amount of design, architecture, and planning
> > that has to go into such a project.  It's not as easy as
> > changing "a DHCP range" or a "handful of RAs" even if they
> > were Globally unique IPv6 (or globally unique IPv4 addresses).
> > I agree that IPv6 stateless Auto-Config adds a lot of room to
> > make this an easier process but it doesn't solve the unique
> > problems that come up in merging two large enterprise networks.
>
> If they are globally unique, one doesn't HAVE to renumber.
> And indeed the problems of renumbering lie in configuration items
> not in the network numbering itself.
>
> > * Yes.  There is the chance that SL will "overlap" in a
> > situation of merging two IPv6 enterprise networks.
> > But if we don't offer SL.  What will people
> > choose for private addressing?  I would REALLY like to see an
> > alternative to SL that maybe provides for a set of bits that
> > relate to Geography?  ISP?
> > etc?  I'm open to other suggestions.  But without an
> > alternative I have to vote "NO".  Let's see some alternatives?
> > Please, again!=> Let's not alienate those that are already
> > afraid of change by "changing" things.
>
> A central registry where one can obtain unique disconnected
> space per /48.
>
> > > > - Other (please specify).  Continually changing the specs
> > will only
> > > > further alienate those that refuse to adopt.  Not offering
> > > > equivalents to what exists today (rfc 1918) will
> > accomplish the same.
> > > > (and  do we really want to put all the NAT vendors out of
> > business?
> > > ;-)
> > > > "Necessity drives ingenuity."
> > >
> > > Here you are already levelling NAT with SL.
> > > If you already intend to NAT, stay with IPv4, you will have the same
> > > problems and still won't get back the end to end connectivity which
> > > was one of the major things IPv6 was built for. Or we could just
> > > do IPv6 with 32bits again...
> >
> > * Yes if anyone chooses to do NAT.  They should have that
> > option, and an IPv6 address space to appropriately use for
> > such functionality [it's a feature, not a bug](some people
> > feel strongly that NAT is a security feature).
>
> If you intend to do it then just stay with IPv4 without
> end-to-end connectivity and don't drag down the rest of us.
>
> > * Again we shouldn't "take away" from what is already there.
> > From what people are comfortable with!  I am not a proponent
> > of NAT (bottlenecking, added administration, etc...)  but we
> > need to continue to support environments that exists today.
> > If a company decides to incorporate NAT in their firewall(s)
>
> "Look mom he said NAT and firewall in one sentence"
>
> > (or even NAT-PT/NAPT-PT), well let's let them!  We've
> > introduced these ideas in IPv4, they exist today.
>
> They didn't pay with money a very long time ago.
> Should we now all go exchange eggs, chicken and wild
> beast again as a currency?
>
> That's called evolution. There is no need for NAT which
> is apparently the sole reason you want SL.
>
> > The current "IPv6 RFC's" and "IPv6 literature" document this
> > continued support clearly (FEC0::/10).
>
> And people have realized that it was a mistake and want to
> clear that mistake up before it starts to hurt hard.
>
> > Constantly changing the standards scares "people" away from
> > adoption.  Do we really want to make such a drastic change.
> > (I'm still trying to deal w/ the deprecation of A6 records.
> > They worked GREAT in renumbering!)
>
> If you have a real use for A6, just like A6, then document
> it and bring it forward, thats where these WG's are all about.
>
> > Yes.  IPv6 offers 340 undecillion (or sextillion depending
> > on where your from) addresses.  Enough to address every
> > device in the world.
>
> Indeed so where did we need SL for again? To NAT????
>
> > Great.  But we need to offer all the functionality that
> > exists today.  They exist for reason (besides just lack
> > of address space.
>
> And which reasons where that again, I don't recall you
> mentioning ANY reason here that doesn't relate to NAT.
> Again, if you want NAT, then stick to IPv4.
>
> >  Some people feel secure/comfortable behind NAT!)
> >  ...my $.02
>
> You should not feel secure behind a NAT. NAT has nothing
> to do with security. Not routing a prefix somehow has though.
>
> Apparently you have no other need for SL than to NAT.
> I rest my case.
>
> Greets,
>  Jeroen

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to