[pfx] Re: MTA-STS without DNSSEC - does it do anything (for google)

Viktor Dukhovni via Postfix-users <[email protected]>
Newsgroups gmane.mail.postfix.user
Message-ID <[email protected]>
On Wed, Apr 01, 2026 at 01:56:10AM -0400, Curtis Villamizar wrote:

> > At the scale of Gmail, and given Google's data storage and retrieval
> > capacity, MTA-STS policy storage is negligible noise not even worth a
> > moment's thought.
> 
> Lookup speed can be an issue if storing that amount of policy.  Best
> lookup is order log N so order log N disk reads.  For 400M domains if
> you can fit 256 in an 8KB block (storing only domain name and an
> index, that's roughly 1M blocks so roughly 16 disk reads per lookup
> with less due to caching.  Another way is to hash the domain name and
> do a lookup on the hash value, not requiring long names or variable
> length names in the lookup.  A short hash with possible collisions
> would make comparisons easier and still work since collisions could be
> resolved.  Of course that is 8GB of search storage so could be fit in
> RAM these days.  I suppose with limited MTA-STS these days it is not
> yet an issue.

Google's storage SREs and distributed database designers have solved
this problem on a much larger scale than required for MTA-STS policy
lookup.

> > No downgrade resistance during policy refresh, but if policy refresh
> > happens before expiration, and is retried multiple times on error,
> > perhaps some resilience could be possible, but this takes effort
> > to get "right" and to prioritise downgrade resistance in the first
> > place.
> 
> So even if no mail is sent for a while policy is refreshed?

Yes, if practical for a smaller site.  Some might say that if Gmail
rarely sends you email, you don't really count as a mail recipient.
I'm not endorsing that view...

> Thanks for the public service announcement.  I've managed to mess this
> up when generating new keys so a good reminder.  I've been guilty of
> sloppy rollover practice, but getting better at it.

It is, in my view, better to not deploy DANE, than to deploy it so
poorly that it is broken a non-trivial fraction of the time, and
does not get fixed without external reminders.

Currently, out ~4.3M domains with DANE SMTP TLSA records for
their MX hosts, ~14k have at least one broken MX host... :-(

-- 
    Viktor.  🇺🇦 Слава Україні!
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.