Re: [v4tov6transition] Any Experience with Using Behave'sStatelessNAT-PT for IMS-SIP VoIP Application...

["Hui Deng" <[e-mail]>]

Inline please,

> -----Original Message-----
> From: Cameron Byrne [mailto:[e-mail]]
> Sent: Thursday, September 23, 2010 12:23 AM
> To: Hui Deng
> Cc: Mosley, Leonard; [e-mail]; [e-mail];
> [e-mail]
> Subject: Re: [v4tov6transition] Any Experience with Using
> Behave'sStatelessNAT-PT for IMS-SIP VoIP Application...
> 
> On Wed, Sep 22, 2010 at 9:01 AM, Hui Deng <[e-mail]>
> wrote:
> > I guess that NATPT and B2BUA are different story?
> 
> Right, B2BUA is a proxy that terminates and re-initiates the call leg.
>  B2BUA are very specific to SIP and therefore can be very robust in
> their handling of SIP traffic.
> 
> As where any form of NAT is fundamentally just a "network layer"
> function with some ALG to support.  And, ALGs are notoriously poorly
> implemented and seldom standardized.
I think we are inline here, the reason I ask here is because I feel SIP is
not rely 
On NATPT for operator service, but he may ask about Internet SIP service.

> 
> > What about ur opinion about performance of deprecated NAT-PT regarding
to
> > SIP-SDP-RTP?
> >
> 
> I do not recommend anyone to have a going forward network strategy
> based on a deprecated protocol, if it can be avoided.
> 
> VoIP is generally considered very important traffic (billable minutes,
> emergency services, branded services) while Internet is considered not
> very important (best effort,  ...).  That said, i believe most network
> operators feel more comfortable keeping the protocol translation
> infrastructure for VoIP / IMS separate (use a B2BUA or P-CSCF
> functions) from the general use protocol translation (NAT64).  The
> basic logic is that the NAT64, like NAT44 today, will have a lot of
> entropy from the various different types of protocols and
> interactions, as where the B2BUA will be much more focused functions
> with stricter rules and less entropy that can trigger "unforeseen
> feature interactions", aka bugs.
> 
> Also, given my limited scope, i have not seen a strong use case, IMHO,
> for stateless translation since it requires 1 to 1 mapping of IPv4 to
> IPv6, and thus does not solve the address exhaustion issue .... which
> is why folks are doing IPv6 in the first place.
> 
> Generally about NAT64 performance, we expect it to be approximately
> consistent with NAT44 CGN performance, slightly less on some
> platforms.  ALGs generally decrease performance since they require
> more complex logic deeper in the packet.
For Internet SIP, does NAT64 shows the same performance as NAT44, 
We are discussing here. Mosley may have testing experience about how
NATPT is weak about Internet SIP?

-Hui


> 
> Cameron