Re: [TLS] DSS with other than SHA-1 algorithms

[Juho Vähä-Herttua <[e-mail]>]

On May 10, 2011, at 7:06 PM, Nikos Mavrogiannopoulos wrote:
>>> but it makes sense to only assign ciphersuites to algorithm sets that require
>>> ECDSA certificates because that is the common case?
>> It's the common case based on a survey of all (known) publicly-deployed
>> servers running TLS-ECC, representing (at least) Microsoft, Apache (presumably
>> OpenSSL), and NSS.  That's the three biggest players in the market (and some
>> smaller ones as well), so I think it's a pretty good representative sample.
> 
> I don't really understand. So is ECDSA used for self-signed
> certificates? Is it used for client certificates (and thus not
> discovered by this study?). According to me this study would be an
> argument for the _RSA ecc ciphersuites rather than the ECDSA.

I was about to say the same. The draft refers to http://msdn.microsoft.com/en-us/library/aa374757%28v=vs.85%29.aspx as a major implementation already doing the suggested suites, but at the same time drops all ECDHE_RSA ciphersuites from the exact same document and only includes the ECDHE_ECDSA ciphersuites. If it's true that there's basically zero ECDSA certs signed by a trusted CA, it would be a quite strong argument to do the opposite.

What's the motivation of dropping ECDHE_RSA suites if there are millions of RSA certificates?


Juho