> On Jan 2, 2018, at 7:03 AM, Ranjana Mukhia <[email protected]> wrote:
>
> 1.If we are going to implement MTA-STS then, whether it should be compulsorily
> used with DANE?
Nothing is compulsory. The standards will tell you how to use STS, but
cannot compel its use. Implementations that strive to avoid downgrade
and MiTM attacks should generally not permit weaker policies to downgrade
concurrent stronger policies. Therefore, if:
1. The domain's MX records are DNSSEC validated, and
2. The MX hostname to which a connection is established
is also in a signed zone (its A/AAAA records are signed,
or result from a CNAME alias and the initial CNAME
from the MX hostname is secure), and
3. Secure TLSA records are published at the domain obtained
by prefixing _<port>._tcp. (port is typically "25") to
either:
a. The secure full CNAME expansion of the MX hostname,
or else if not secure, or, if no TLSA records present
there (NXDomain or NODATA, abort use of MX host on
TLSA lookup failure), at
b. The original MX hostname (abort use of MX host on
TLSA lookup failure).
then, a sender that supports and enables DANE should typically ignore STS.
Requiring both when both are published feels too fragile to me, and requiring
either downgrades DANE security.
On the other hand, some domains might have only a partial DANE
implementation, where some MX hosts have TLSA records and others
do not. In that case, when a given MX host is not secured with
DANE, but the domain has STS policy, it makes sense to apply STS
when delivering via that MX host (assuming STS is supported and
enabled at the sender).
> 2.Whether MTA-STS is fully capable of securing email transmission
> without the help of any other technologies like DKIM,SPF,DMARC or DANE?
You have not defined "securing email transmission". STS and DANE
are both designed to authenticate the nexthop SMTP destination to
the sending MTA, and to transmit the message envelope and body over
a TLS channel that provides integrity and confidentiality of the
transmitted data.
STS does this subject to the integrity of WebPKI certificate
issuance and potential downgrade on first-contact (and possibly
after previous cached policy has expired if email flow to the
nexthop domain is infrequent). Given that DV certificates are
issued based on "domain control", the WebPKI can be at most as
secure as the DNS.
DANE does this subject to the integrity of DNSSEC for the
domain and its ancestor domains up to the trust-anchor used
(typically the DNS root zone keys). DANE does not have a
first-contact security gap.
Neither DANE nor STS provide message origin authenticity,
these are hop-by-hop security mechanisms that authenticate
only the receiving system, not the sender.
--
Viktor.
_______________________________________________
Uta mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/uta
