Should we name the next version of TLS 1.3, TLS 4, or something else?
Work on the next version of TLS is wrapping up and it’s expected to be finalized in early 2017. What’s left is mainly small decisions about fine-tuning the protocol’s design. But one major thing that has not been decided on is the name.
Since the first draft was published, it was assumed that this next version of TLS would be named TLS 1.3. The current version is TLS 1.2, and all the TLS versions have followed .1 increments since TLS 1.0 was released in 1999.
So this should be easy, right? Just name it TLS 1.3 and be done with it.
But there is disagreement within the community, which predates the release of TLS 1.0.
As you may know, the protocol was originally named Secure Sockets Layer, or SSL, by Netscape, who originally invented it. The name was changed to TLS when the Internet Engineer Task Force (IETF) took over development, mainly to satisfy egos.
The full version history goes like this: SSL 2.0, SSL 3.0, TLS 1.0, TLS 1.1, TLS 1.2.
The jump from SSL 3.0 to TLS 1.0 has always been cited as a source of confusion. For most system administrators and webmasters, keeping up with the release histories of every protocol isn’t on the top of their to-do list. So they assume that SSL 3.0 is newer because it has the highest version number, and enable that version (in the worst case scenario, they enable it instead of TLS 1.2).
This confusion is made worse by the fact that “SSL” still has more name recognition and is more widely used even though it is no longer the protocol’s official name. This is partially our own industry’s fault, as many major players have kept the SSL name, such as OpenSSL (the most widely used SSL/TLS library) and even our own website.
So now that the next version of TLS is almost out, some in the community want a chance to fix the confusion. Two popular alternate names to TLS 1.3 have been proposed: “TLS 4” and “TLS 2017.” Both names hope to address the version number conflict and give the newest version the biggest number.
The surprising part is that there are pretty convincing arguments for all three.
Sticking with TLS 1.3 ‘rocks the boat’ the least, and has lots of support for that very reason. While it’s likely that TLS 1.3 will continue the same version confusion, more sensible security settings make this less of an issue than it was a decade ago. For instance, because all the major web browsers are fairly proactive in disabling insecure options (like SSL 3.0) the server’s configuration is no longer as important. New versions of server software are unlikely to even have SSL 3.0 compatibility, so this versioning issue will sort itself out once SSL 3.0 is no longer around.
Changing the name to TLS 4 would make it clear that this is the latest version, but throws out any numerical hierarchy since “4” would be the best, “3” is bad, and “1.3” is good. TLS 2017 makes it expressly clear when the version was released, and there is some precedent for this naming convention.
No matter what the final name is, TLS X, as we will call it, for now, will bring great improvements to the protocol. A lot of the built up cruft – like the truly insane number of encryption ciphers currently available – will be removed, greatly simplifying configuration and lessening the risk of an insecure configuration. Zero round-trip connections (abbreviated as 0RTT) will enable lightning-quick handshakes.
What would you name it?