--On Monday, 07 January, 2008 22:35 -0500 Dave Crocker
<dhc2@xxxxxxxxxxxx> wrote:
>
>
> Tony Finch wrote:
>> On Fri, 4 Jan 2008, John C Klensin wrote:
>>> On Friday, 04 January, 2008 12:01 -0800 Bill Manning
>>> <bmanning@xxxxxxx> wrote:
>>>> The general answer when needing to communicate between
>>>> similar applications that run on different address families
>>>> has traditionally been the application layer gateway (ALG)
>>>> ...
>>
>> MTAs *are* ALGs.
>
>
> (Uh Oh.)
>
> Nope.
>
> MTAs are email relays. ALGs can perform syntactic and
> semantic transformations. A simple relay can't. ALGs connect
> similar applications that entail heterogeneous technologies,
> conventions, whatever.
>
> Now it happens that email gateways have to include MTA
> functionality, since they all operate as relays, also. And
> lots of software the is email gateway-capable are often used
> merely as MTAs, nicely confusing the heck out of us.
I'd make a further distinction and claim that even many email
gateways don't rise to the level (or sink to the depths) of what
we usually think about as ALGs. First of all, I would contend
that an SMTP system that accepts an SMTP-and-MIME conformant
message that arrives over an IPv6-based TCP connection and that
is sent back out, using an unchanged address, over an IPv4-based
TCP connection is performing a standard relay function, not even
a mail gateway one (I think that is consistent with Dave's
comment above). But even a mail gateway that, e.g., needs to do
some rewriting of addresses to get from the systems on one side
to the systems on the other still isn't in the ALG business as
I, at least, usually understand it. The transformations are
mechanical, they don't require any external information, and
they don't inherently depend on the underlying transport
technology. So I would suggest that some, but not all, email
gateways are ALGs... and that no relay is.
> But architecturally, ALGs are seriously different. Stated
> simplistically, an MTA performs very few functions and makes
> minimal changes to a message, where the 'changes' are mostly
> some additions. An email gateway can massively alter the
> object. It has permission to do whatever is necessary to make
>
> d/
>
> ps. No one will be surprised that I've documented this issue
> in the email
On the other hand, they might be surprised that we seem to be
agreeing <grin>
john
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf