You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
SRP is a PAKE, the main benefit of a PAKE is that the password gets used as part of the key exchange protocol, authenticating both the client and the server simultaneously. TLS-SRP combines SRP with TLS, and can be deployed today.
More specifically, the benefits are:
An active attacker (such as may happen with a self-signed server certificate) doesn't get any information about the password. Not the password itself (as with SASL PLAIN), not the password hash (as with SCRAM). It is completely zero-knowledge to an active attacker.
An active attacker will cause the TLS connection to be closed. The key exchange will fail, and without a successful key exchange, there's no TLS connection.
It is highly beneficial for a server to properly hash its passwords when using TLS-SRP. Encouraging TLS-SRP support would also encourage proper password handling.
TLS-SRP does have some drawbacks, tho: it has been broken quite a few times in the past, and is up to version 6a nowadays(!). However, I strongly feel like the benefits far outweigh the risks. Additionally, we can deprecate TLS-SRP when TLS-OPAQUE comes out.
many networks support SASL EXTERNAL. whether they'll take SRP or CertFP is another story, and not defined by any RFCs. (neither IRCv3 nor SASL EXTERNAL define it.)
I strongly recommend supporting TLS-SRP with SASL EXTERNAL. The existing SASL EXTERNAL should then be renamed to SASL EXTERNAL (Client Cert).
https://tools.ietf.org/html/rfc5054
SRP is a PAKE, the main benefit of a PAKE is that the password gets used as part of the key exchange protocol, authenticating both the client and the server simultaneously. TLS-SRP combines SRP with TLS, and can be deployed today.
More specifically, the benefits are:
TLS-SRP does have some drawbacks, tho: it has been broken quite a few times in the past, and is up to version 6a nowadays(!). However, I strongly feel like the benefits far outweigh the risks. Additionally, we can deprecate TLS-SRP when TLS-OPAQUE comes out.
See also https://blog.cryptographyengineering.com/2018/10/19/lets-talk-about-pake/
The text was updated successfully, but these errors were encountered: