Comments draft-palet-v6ops-proto41-nat-01.txt

[mariana.nikolova@philips.com]

Hi Jordi,

my "last minutes" comments on your draft. Hope they are not "too late" comments. 
In general, I am very positive over the document. The remarks are mainly for extra clarification!

General comment: 

1)        "NAT boxes" is very often used in the draft meaning actually "IPv4-only NAT node". An IPv6/IPv4 node that is a 6to4 router implements also NAT only for IPv4 packets. Then, it can be referred to as a NAT boxes as well. However, no need to implement proto41 on this box (although it can be and you mention this in the draft but then it's overdone and rather confusing for the router producer what (6to4 or proto41) and when (IPv4-only or IPv6/IPv4 node) to implement ). The strong need for support of proto41 is in "IPv4-only NAT nodes" in order to support IPv6 tunnels. That is why I recommend the use of "IPv4-only NAT nodes" instead of "NAT boxes". Further see my remark in section 5.
2)        Very often in the draft "this behavior" is used meaning "forwarding protocol 41" or the nickname "proto41". Replacing the former with the latter is more tangible and I recommend it.

Specific comments: 

- Abstract
I agree with the Rute's comment here

1. Introduction
1-2 paragraphs: Substitute with a short intro to the scene, e.g.  

Nowadays homogeneous IPv4 private networks are massively deployed. The first migration step from IPv4 to IPv6 for these networks is adding an IPv6 node (or cluster of IPv6 nodes). In this case the IPv6 node communicates with IPv6 peers in the public Internet via IPv6 packets tunneled into IPv4 ones. Most of the existing solutions suppose that the router of the private network is begin-/end-point of the tunnels. Typical examples of such routers are 6to4 IPv6/IPv4 routers that have already been commercially deployed. However, nowadays most of users have IPv4-only NAT routers in their private networks and they are not willing immediately to invest in a new router for whatever reasons, mostly financial.
That is why in this draft we propose to make use of protocol 41 forwarding in IPv4-only NAT routers. This mechanism allows IPv6 tunnels and therefore facilitates the migration path from IPv4 to IPv6.  

4 paragraph: revise, for example
The scenario illustrated above has been tested with several IPv4-only NAT boxes/routers that have successfully established IPv6 tunnels between tunnel clients in a private network and tunnel servers in the public Internet. In the test we have used three well-known Tunnel Broker implementations (BT, Freenet6 and TILAB) and routers from 6Bone, Consulintel, Euro6IX and UPM networks.

5 paragraph: This can -->  This scenario can ..

2. Rationale for this behavior  --> Rationale for using protocol 41 forwarding

3. Behavior of different NAT types
I do not understand Rute's confusion here since Basic (or Traditional) NAT, NAPT and Bidirectional (or two-way) NAT is commonly used terminology. 

5 paragraph in 3.2: This can also be combined with basic NAT.  -->This port forwarding is often combined with basic NAT leading to NAPT.

2 paragraph in 3.3: Revision required. What you wanna say is written so compact that it is rather getting non-understandable.

1 paragraph in 3.4: N.B. At the beginning of this section make clear that "configurable" NAT can be either 3.1 or 3.2 or 3.3. Each of the first three types can act as a fully bidirectional NAT for 41-packets if it is configurable. 

4. Applicability

1 paragraph: 
inside-to-outside sessions  outgoing sessions, outside-to-inside sessions  incoming sessions
So, this is consistent with the terminology you use later.
If my first general comment is applied here the Rute's remark is resolved. 

6 paragraph: the application of this procedure  --> the application of 41-packets forwarding 

7 paragraph: move to Introduction and consider language revision

8 paragraph: move to Introduction after the intro paragraph I suggested.


5. NAT design consideration  --> NAT design consideration and recommendations

I agree with the Rute's first comment here. 

3 paragraph: confusing the reader and contradicting the general idea. I would rather say something like 

The proto41 and 6to4 are complementary transition mechanisms that facilitate the migration from IPv4 to IPv6.These two mechanisms are in general applicable in different migration steps. The proto41 mechanism is mostly relevant and applicable for IPv4-only NAT routers whereas 6to4 mechanism is mostly relevant and applicable for IPv6/IPv4 routers. Notice that an IPv6/IPv4 router that is 6to4 enabled and implements NAT only for IPv4 packets does not need to be proto41 enabled. 

Some extra conclusions that you may add to make this section complete:

·        The proto41 adds an enhanced feature to the IPv4-only NAT routers, namely enabling them to forward IPv6 packets encapsulated into IPv4 ones, i.e., allowing for IPv6 tunneling.
·        The proto41 completely preserves the users investments in the existing IPv4 networks. This is essential for gaining market acceptance.
·        The proto41 allows for gradual migration from IPv4 to IPv6 networks making the migration path easy and acceptable for the users.  


Greetings, 
Mariana
-----------------------------------------------------------------------------------------------------------------
Dr. Mariana Nikolova    
Philips Research Laboratories Eindhoven (IST/SwA/DS)
Prof. Holstlaan 4, 5656 AA, Eindhoven, The Netherlands
room: WDC 1.35,     phone: +31-40-27-45455
e-mail: mariana.nikolova@philips.com 
-----------------------------------------------------------------------------------------------------------------