Hi there,
New applications can do this themselves, but what about old applications such 
as TFTP?

Regards,
Ewan 



yug...@telincn.com
 
From: otroan
Date: 2017-05-26 16:52
To: yugang
CC: vpp-dev
Subject: Re: [vpp-dev] ALG
Hi there,
 
> Many applications have two channels, one is data channel, the other is 
> control channel, both have different ports,
> so we need ALG to help these protocols to cut through FW, i think it is a 
> pretty important feature and
> it has been surpported by almost every  manufacturer on their NAT devices.
 
It has been supported with varying success.
Take RFC5389 and the XOR-MAPPED-ADDRESS, where the STUN protocol where forced 
to hide IP addresses in the payload because misguided ALGs rewrote them.
 
ALGs are often more of a hinderance to application developers than not. ALGs 
are hard to get right, and they lead to ossification.
It is easier to write application if you have a predictable network. Any 
application on todays network either uses port 443 or it has to deal with NAT 
traversal.
 
The IPv4 Internet wth address exhaustion is evolving. Address sharing is a 
fundamental part of the architecture now. Take mechanisms like MAP-E (RFC7597) 
where ALGs will not be included.
 
If you have a specific request for a particular application we can talk about 
it. But in general my stance is "not the network's problem to solve".
 
Best regards,
Ole
 
_______________________________________________
vpp-dev mailing list
vpp-dev@lists.fd.io
https://lists.fd.io/mailman/listinfo/vpp-dev

Reply via email to