ietf
[Top] [All Lists]

Re: Internet Technology Adoption and Transition

2014-04-24 10:18:56
Hi Ned,
At 19:53 22-04-2014, Ned Freed wrote:
I'm not sure what the report means by "evolve", but by any reasonable
definition, I believe SMTP continues to evolve.

Ok.

If you're talking about standards work, the IESG just approved the RRVS
document which specifies a new SMTP extension. In recent times we've approved
other extensions such as MT-PRIORITY. And let's not forget about EAI, which is
a pretty major evolution of the protocol.

And this process continues. I rather expect we'll see changes in the STARTTLS space in the not too distant future.

If you're talking about implementations, the first thing to note is that
given the relatively small number of implementations in wide use, a change
to even one or two of them is significant. And while progess is slow,
it's pretty constant from my perspective.

I was looking at it in terms of what is implemented and what is deployed. There has been standards work; some of examples are mentioned above. I agree that EAI is a pretty major evolution of the protocol.

As for actual deployment, things are even more skewed, which makes them
difficult to measure. For example, if a major wireless vendor implements the
BINARY extension in both their client and their own server, that can affect a
huge swath of traffic without any need for anyone else to change.

There are different ways to look at usage. As an example, BINARY may be about 25% of what's deployed (please treat that figure as unverified).

The deployment of older extensions also changes over time, and it doesn't
always increase. I'm fairly sure I'm seeing more of ENHANCEDSTATUSCODES but
less of NOTARY. (And yeah, I know given the relationship between the two that's
wierd. I'm just reporing what I've observed.)

The statistics I looked at show a little more of ENHANCEDSTATUSCODES.

Moreover, there can be evolutionary changes having nothing to do with
extensions per se. For example, it seems that not only is use of STARTTLS on
the rise, what cipher suites are enabled seems to be changing, probably because
a lot of sites are actually paying attention to such details when prior to the
Snoden thing they care.

I haven't look into the STARTTLS statistics recently to be able to compare the figures with the ones before the Snow thing.

Regards,
S. Moonesamy