Re: [pcp] clarification on Axternal Address assignment [was RE: pcp-base-19]

[Tina TSOU <[e-mail]>]

-----Original Message-----
From: Dan Wing [mailto:[e-mail]] 
Sent: Friday, December 23, 2011 11:02 AM
To: Tina TSOU
Cc: [e-mail]; [e-mail]
Subject: clarification on Axternal Address assignment [was RE: [pcp] pcp-base-19]

> -----Original Message-----
> From: Tina TSOU [mailto:[e-mail]]
> Sent: Friday, December 23, 2011 12:52 AM
> To: Dan Wing
> Cc: [e-mail]
> Subject: RE: [pcp] pcp-base-19
> 
> Dan,
> In line...
> 
> - Tina
> 
> -----Original Message-----
> From: Dan Wing [mailto:[e-mail]]
> Sent: Wednesday, December 21, 2011 3:56 PM
> To: Tina TSOU
> Cc: [e-mail]
> Subject: RE: [pcp] pcp-base-19
> 
> > -----Original Message-----
> > From: Tina TSOU [mailto:[e-mail]]
> > Sent: Wednesday, December 21, 2011 3:06 PM
> > To: Dan Wing
> > Cc: [e-mail]
> > Subject: Re: [pcp] pcp-base-19
> ...
> > 		2. P28, section 9
> >
> >
> > 		"It is REQUIRED that the PCP-controlled device assign the
> > same
> > 		external IP address to PCP-created explicit dynamic
> > mappings and to
> > 		implicit dynamic mappings for a given Internal Address. In
> > the absence
> > 		of a PCP option indicating otherwise, it is REQUIRED that
> > all
> > 		PCP-created explicit dynamic mappings be assigned the same
> > external
> > 		IP address."
> >
> >
> >
> > 		How about replace "In the absence of a PCP option
> > indicating otherwise,
> > 		it is REQUIRED that all PCP-created explicit dynamic
> > mappings be
> > 		assigned the same external IP address."
> >
> > 		with
> >
> > 		"It is indicated by the PCP client that PCP-created
> > explicit dynamic
> > 		mappings be assigned the same external IP address, unless
> > there are
> > 		explicit reasons of not doing so, e.g.
> > 		http://tools.ietf.org/html/draft-penno-pcp-zones-00"?
> >
> >
> > 		Because "It is REQUIRED that the PCP" give the requirement
> > from the
> >
> >
> > 		server's point of view, I think we should also give the
> > requirement
> >
> >
> > 		from the client's point of view.
> >
> >
> > 		This is more or less what Dan suggested before, perhaps an
> > oversight.
> >
> >
> >
> > 	I don't understand the nuance between the wording.  Can you give
> > an
> > 	example of what the existing text prohibits / breaks / disallows?
> >
> > The existing text disallows client to request same external IP
> > proactively.
> 
> 
> The client can request the address it wants in the Suggested
> External Address field, which exists for both MAP and PEER.
> [Tina: Excellent! So some descriptive text is needed here to describe
> these protocol operations before digging into the messages and fields.]
> 
> > You agreed my comments earlier before.
> 
> I recall a discussion on this previously, yes.  I don't recall any
> definitive conclusion.  Here are pointers to the end of two
> threads related to the previous discussion:
> 
> http://www.ietf.org/mail-archive/web/pcp/current/msg01632.html
> http://www.ietf.org/mail-archive/web/pcp/current/msg01627.html
> 
> 
> If there was a definitive conclusion previously please provide
> a pointer and accept my apologies for my oversight.
> 
> [Tina:
> http://www.ietf.org/mail-archive/web/pcp/current/msg01391.html
> You said in Date: Tue, 7 Jun 2011 11:26:39 -0700
> "I think see what you're wanting the document to say.
> 
> How about this proposed text, instead:
> 
> 
>   If there is already an active explicit dynamic mapping,
>   it will be mapped to a certain external IP address.
>   When the PCP client makes another explicit dynamic mapping,
>   it SHOULD place the external IP address of the existing
>   mapping into the Requested External Address of the MAP
>   request.  By doing this, all of the PCP client's explicit
>   dynamic mappings will be on the same external address.
> 
> 
> Does that say the same thing?
> 
> Is it acceptable?
> 
> -d"
> 
> I answer now "Yes, it is." Sorry for the very laaate answer from summer
> till X'mas. Merry X'mas!
> ]

Ok, I think I understand now.  My proposed text, above, isn't
quite right, either (the last sentence of my proposed text is
cannot be true, because no matter the value in Suggested 
External Address, the PCP server can ignore that value; it is
only draft-penno-pcp-zones that changes that).


Please review this change:

OLD (contents of -19):
   It is REQUIRED that the PCP-controlled device assign the same
   external IP address to PCP-created explicit dynamic mappings and to
   implicit dynamic mappings for a given Internal Address.  In the
   absence of a PCP option indicating otherwise, it is REQUIRED that all
   PCP-created explicit dynamic mappings be assigned the same external
   IP address.  It is RECOMMENDED that static mappings for that Internal
   Address (e.g., those created by a command-line interface on the PCP
   server or PCP-controlled device) also be assigned to the same IP
   address.  Once an Internal Address has no implicit dynamic mappings
   and no explicit dynamic mappings in the PCP-controlled device, a
   subsequent PCP request for that Internal Address MAY be assigned to a
   different External Address.  Generally, this re-assignment would
   occur when a CGN device is load balancing newly-seen hosts to its
   public IPv4 address pool.

NEW:
   The MAP and PEER requests include a Suggested External IP Address
   field.  This field is a hint and ignored by the PCP-controlled device
   if there are other mappings for that Internal Address, unless there
   is an explicit Option in the request indicating otherwise (e.g.,
   PREFER_FAILURE, [I-D.penno-pcp-zones]).  For a PCP explicit dynamic
   mapping request for a given Internal Address, which does not
   contain an explicit Option to override the following behavior, the
   PCP-controlled device MUST assign the same External Address to that
   PCP-created explicit dynamic mapping and MUST assign it the same
   External Address as existing implicit dynamic mappings.  Static 
   mappings for that Internal Address (e.g., those created by a 
   command-line interface on the PCP server or PCP-controlled device) 
   SHOULD also be assigned to the same External Address.  Once an 
   Internal Address has no implicit dynamic mappings and no explicit 
   dynamic mappings in the PCP-controlled device, a subsequent implicit 
   or explicit mapping for that Internal Address MAY be assigned to a 
   different External Address.  Generally, this re-assignment would 
   occur when a CGN device is load balancing newly-seen Internal 
   Addresses to its public pool of External Addresses.

[Tina:
>From the client point of view, simpler and more straight forward, how about this?
"The client should request same external IP address if there is any existing mapping, unless there is an explicit reason for not doing so, e.g. [I-D.penno-pcp-zones]]."
]

-d