[BEHAVE] need for ALGs in NAT64 scenarios [was RE: adopt draft-van-beijnum-behave-ftp64-06 as WG item]

> -----Original Message-----
> From: behave-bounces@ietf.org 
> [mailto:behave-bounces@ietf.org] On Behalf Of bo zhou
> Sent: Wednesday, October 28, 2009 5:00 AM
> To: Iljitsch van Beijnum
> Cc: behave@ietf.org WG; Zhen Cao
> Subject: Re: [BEHAVE] adopt draft-van-beijnum-behave-ftp64-06 
> as WG item
> 
> Hi Iljitsch,
>  
> "minimize" does not mean avoid in my mind. 
> BTW, you ask "which protocols do you feel need ALGs to work 
> through 6-to-4 translator?"
> I guess Zhen want to answer your question using RTSP, SIP as 
> two examples.
> I believe more applications rather than RTSP, SIP need to do 
> ALG in such scenario, but it is really depend on how the user 
> use these applications. for example, a standard HTTP protocol 
> can include IP address or domain name inside. if the user 
> choose the IP address filling, then the ALG cannot avoid.

An ALG is not necessary for that problem.  I describe a workaround
for that specific problem in
http://tools.ietf.org/html/draft-wing-behave-http-ip-address-literals-01
which does not use an ALG.  However, as noted that draft, my workaround 
may not properly handle Java or Javascript which contains IPv4 
address literals -- but neither will an ALG (unless the ALG is
going to interpret Java and Javascript, which *significantly* 
increases its complexity and CPU hit to solve a small problem; 
see below).

> But likely, more user use domain name rather than IP address for 
> HTTP application which need not ALG in the translator.

Yes.  I analyzed the likelyhood by looking at the top 1,000,000 
domains per alexa.com, results are at:
http://tools.ietf.org/html/draft-wing-behave-http-ip-address-literals-01#appen
dix-B
which says:

...
   An examination of Alexa's top 1 million domains [Alexa] at the end of
   August, 2009, showed 2.38% of the HTML in their home pages contained
   IPv4 address literals.  
...
   Of the top 1 million websites at the end of August, 2009, 3455
   (0.35%) of them are IPv4 address literals (e.g., http://192.0.2.1).
...

-d

> regards,
>  
> Bo 
> 
> 
>  
> On Wed, Oct 28, 2009 at 6:29 PM, Iljitsch van Beijnum 
> <iljitsch@muada.com> wrote:
> 
> 
> 	On 27 okt 2009, at 16:59, Zhen Cao wrote:
> 	
> 	
> 
> 			Which protocols do you feel need ALGs 
> to work through IPv6-to-IPv4
> 			translators?
> 			
> 
> 
> 		RTSP, SIP where there needs SDP to carry the address.
> 		
> 
> 
> 	I'm not sure about the status of RTSP, but I believe 
> that it's used less and less exactly because it doesn't work 
> through NAT44 without an ALG...
> 	
> 	As for SIP, I believe the consensus is that ALGs create 
> more harm than benefit.
> 	
> 	The wg consensus is that the use of ALGs should be 
> minimized. I don't believe there would be consensus to 
> specify RTSP and SIP ALGs for NAT64. 
> 
> 	_______________________________________________
> 	Behave mailing list
> 	Behave@ietf.org
> 	https://www.ietf.org/mailman/listinfo/behave
> 	
> 
> 
> 
> 
> -- 
> Regards,
> 
> Bo Zhou
> China Mobile
> 
> 