Podcast
Root Causes 284: 90-day SSL Certificates Are on the Way


Hosted by
Tim Callan
Chief Compliance Officer
Jason Soroko
Fellow
Original broadcast date
March 10, 2023
The Google Chrome root program recently announced its intention to reduce the maximum term for public SSL certificates to 90 days. In this episode we explain this announcement and its implications and speculate on timing for this reduction.
Podcast Transcript
Lightly edited for flow and brevity.
Recently, we saw an update – and I’m gonna focus on a little bit of this page with significantly more material than just what we will focus on – but there’s a section here called, Encouraging Modern Infrastructures and Agility. I’ll sample from that and read just a little of it, ok? There’s more to it than this, but here’s the passage:
“In a future policy update or CA/Browser Forum Ballot Proposal, we intend to introduce:
A reduction of TLS server authentication subscriber certificate maximum validity from 398 days to 90 days.”
In other words, what Google has just said is they intend to reduce the maximum validity of an allowable public SSL certificate to 90 days.
So, the first thing, I’m gonna go back to the beginning where it says, “In a future policy update or CA/Browser Forum Ballot Proposal, we intend to introduce:” Now let me break that down because that actually is meaningful. A CA/Browser Forum Ballot Proposal. In other words, we will introduce a ballot that will say the maximum term will be 90 days. We will take that to the CA/Browser Forum. As a little reminder for those of you who haven’t been following the subtleties of all of these things or listening to every single episode of this podcast, there was a process whereby on two occasions ballots were introduced to the CA/Browser Forum to reduce the term of a certificate to one year of an SSL Certificate and both of those ballots were voted down by the majority of the CAs in the CA/Browser Forum. And even in the second ballot, if my memory serves, every single browser consumer voted for the ballot. The ballot still passed because the CAs declined to put it through. After which, Apple declared this was going into their root policy and it was a done deal. It was a fait de accompli, after which the CA/Browser Forum passed a ballot.
So, you know, I think what Google is saying here if you read this wording is we are gonna put out a ballot, we are gonna give you guys a chance to vote for it. In the event you don’t vote for it, we are gonna make it into a root program requirement anyway and it won’t really matter if you voted for it. Right? That is reading between the lines. So this signals a couple things.
One, it signals Google’s seriousness that they’ve made up their mind and this isn’t something they are toying with, it’s something that they’ve decided.
The second thing that it signals is that they’d like to do it through a ballot. They’d like to follow the CA/Browser Forum and do it through a ballot and have a good community and have the community agree, but in the event that it doesn't they are willing to flex their muscles.
The last thing that signals to me is that they don’t want this lingering for years. That is something that says we will move deliberately and reasonably rapidly to make this happen. I see all of three things built into that language there, and I think those are important takeaways for all of us to bear in mind.
And as we said on previous podcasts, Tim, any one of the major browsers, if they make a rule change, typically a lot of them will follow suit and certainly who has to follow suit is the CAs because of the fact that no CA can survive not being completely compliant with every single browser. Every single percentile is just so important.
Now the second subtlety that I want to point out is that – and I’m gonna go down to that latter passage that I read: It’s “a reduction of TLS server authentication subscriber certificate maximum validity from 398 days to 90 days.” And as I said – well, think about it this way. Hey, Jay? How many days in a year?
The last thing I’m gonna point out about that is there’s no cushion. So, we went to 398 days was the old “one-year” cert. Why 398 days? Well, it’s actually 13 months if you picked the very longest month, a 31-day month and you add a Leap Year day. Right? That’s how you get to the 398 days. So it’s the very most days that a 13-month timespan can ever be under any circumstances. That’s why Apple picked that, right? We don’t see any cushion time in here at all. It’s not, eh, we want you to do it every 90 days so we are gonna give you 100 days. It’s 90 days. Which means that in reality, if you don’t want to take things down to the last day – which you don’t – you are probably replacing your cert every 80 days, every 75 days, you know, something materially less than that. And so, in reality, even this thing that I’ve been hearing people say: oh well, you only have to do it four times a year. No. You are really probably doing it five times a year
So, what are we looking to get out of this? We are trying to create crypto agility. Or what I like to call certificate agility is a form, a subset of crypto agility. And they are trying to do that. They are trying to say, we’ll be able to move the new cryptography in really quickly. A good example being quantum-resistant certs. Once it’s possible for me to issue a quantum-resistant cert, if there are certs out there that are lasting for a year then it’s gonna take an entire year to get those cycled out. But on the other hand, if all the certs are only 90 days in length, then, you know, not more than 90 days from today 100% of the certs are gonna be supportive or able to support the new algorithms. So, that’s the kind of thing that they are looking for and this isn’t new. This has been in the public discussion from multiple major browsers for years and years. The need for crypto agility, the need to remove error-prone and difficult manual processes, these are deemed to be essential to the ongoing health and security of the web PKI by multiple thinkers including, based on what I see here, the people who run the Google Root Store Program.
Well, what Google is doing here – that’s courageous because there’s risk here and the risk is - - I mean I can see the pushback. Oh my God, we can’t go to 90-day certs because I don’t have a CLM.
This has been announced. This is being telegraphed by Google early in the year of 2023. They’re not doing that because they don’t intend to take action in 2023. We already know because Google has told us that they are going to move forward with this OCSP ballot. This OCSP ballot is gonna be very important and one of the things it’s gonna do is put another motivator in place for automation and shorter-lived certificates. That ballot I predict will pass. I think that ballot will be very popular and after that passes, that will clear the slate for Google to move on to their next initiative. So, I am predicting that sometime this year – I’m gonna say summertime – we see a ballot that proposes 90-day certificates. Now that may not pass. However, we’ve also seen rhetoric from Google that suggests that in the event that that doesn’t pass that they will move forward with a pronouncement anyway.
Let’s look at those two scenarios. Either it passes or it doesn’t. At the end of the day, the effective date probably doesn’t change in a meaningful way. Let’s just pretend it doesn’t pass. Summer of 2023, it fails. Google probably gives it a little bit of respectful time and then they probably come out and say, well, here’s a pronouncement and here is what we are gonna do. We are gonna make this a root store requirement and make it on this date.
Now, they tend to make these pronouncements at the CA/Browser Forum face-to-faces. So, it will probably happen at the fall face-to-face. Usually, these guys give us about a year. Ok, you got a year. Get your systems worked out CAs because in a year this is gonna happen. So, if you assume that happens now we are looking at essentially Quarter 4 of 2024.
So I’m gonna throw out a date of October 15. For me, the over/under on this is that October 15, 2024 there will be a mandatory requirement to move to 90 day certificates. And, I may be wrong but I just walked through the rationale of how I got there and if someone who is on the inside and really knows what the decision is gonna be and wants to state otherwise, obviously, we will trust them but until that occurs, that is my prediction for when this happens. I’m predicting that it’s very likely, damn near a fait de accompli that in that timeframe we will have that requirement.
I think what the listener of this podcast needs to - - your takeaway is, guys, you are not far away from absolutely needing to automate your certificate lifecycle management. Period.
And so, this really is the time to try to understand what is my profile on automation, what do I have to do to get there and what is preventing me from getting there? If you need to secure some budget, this is the time to be looking to secure that budget. If you need to reallocate some roadmap and some resources, this is the time to be looking at reallocating that roadmap and resources. Get going on it, guys. Google deliberately told us this ahead of time so we all had a chance to prepare. If we don’t prepare, that’s not on them. It’s on us.

