Re: [ietf-smtp] [Emailcore] Proposed ESMTP keyword RCPTLIMIT

[John R Levine <[e-mail]>]

It looks like something like this would handle a lot, quite possibly most, 
of the rate limits that receivers are willing to disclose:

   CONNRATE=m/s MSGRATE=m/s

Where m is a number of connections or messages or whatever, and s is a 
number of seconds.

I suppose it could also have something like SCOPE=HOST or SCOPE=GROUP to 
say whether it's for a single host or a group but I don't see any way of 
describing groups that wouldn't be too complicated and I doubt that many 
mail providers would want to tell hostile strangers that much about their 
internal architecture.

R's,
John


On Tue, 20 Apr 2021, Ned Freed wrote:
>>> A little surveying might be in order.
>
>>> I see that Postfix lets you set both the numerator and denominator of
>>> the rate limit, but the denominator defaults to a minute.  It has
>>> separate rate limits for connections and messages.
>
>>> Exim also lets you set both, the examples use an hour.  The
>>> documentation says that if you use a large interval that lets people send bursts.
>
>> Sounds familiar. Oracle Messaging Server lets you set both; the denominator default is one hour.
>
>> We store the limits in memcache or Redis so they can apply across mutiple MTAs.
>> The usual limit is per-IP, but we support per-host and per-domain as well.
>
>> The demoninator in PowerMTA is a string rather than a number, possible values are sec, s, min, m, hr, or h. It supports per-IP and per-host and probably other stuff; my familiarity here is limited.
>
>> I assume the Exim and Postfix limits are per-host, which in most cases translates to per-IP.
>
>> I have to say there's more consistency here than I expected.


Regards,
John Levine, [e-mail], Taughannock Networks, Trumansburg NY
Please consider the environment before reading this e-mail. https://jl.ly