Grim thoughts on the future of 'CLOUD-6999'

CLOUD-6999, which I often talk about with exasperation, is a legendary Jira task, at least among those in Confluence. It's legendary because it hasn't shipped a feature in 12 years, despite saying it's in development.

The Atlassian product family is broadly divided into server, data center, and cloud products. The server suite is used by small to medium-sized organizations to use Atlassian products in an on-premise environment, but it was discontinued a few years ago and was discontinued last year. In my previous experience, some organizations started using the Confluence server product due to security concerns and didn't want to pay additional license fees for each major version upgrade, so they kept the old server version, giving their users (employees) an outdated experience with Confluence. The Data Center Suite is what we use in our on-premises environment when we use Atlassian products at scale, and the minimum contract unit for Confluence starts at 500 users. You're looking at tens of thousands of dollars in annual software fees alone, which is probably more than you can afford to spend on a data center environment. While sales and support for the server version have ended, the data center version is still available and continues to be developed and supported.

Finally, the cloud version is the use of Atlassian products in an environment that Atlassian has deployed directly to AWS, which is a relatively less secure and sometimes unreliable environment compared to the data center version. Despite the relative lack of security, in recent years Atlassian has focused on developing environments that meet the various compliance requirements on its roadmap.

CLOUD-6999 is registered in July 2011 with the requirement to allow the use of custom domains, especially in confluence cloud environments. This task has been neglected for 12 years now, and the PM to whom it was assigned a few years ago has been popping up periodically to update us on progress. However, I have doubts about whether this progress is being made because I don't know if it's because I don't read English very well, but there's nothing there, just a statement that custom domains are important to us and that we're doing various preparatory work for them. At some point, people who got tired of waiting after receiving multiple of these seemingly fruitless replies over the years started making and selling t-shirts and stickers, and I bought one and put it on my laptop and iPad to remind me every day of the absurdity of this 12-year-old task. Others have gone crazy waiting and started playing tic-tac-toe with the replies, and it's gotten to the point where I'm seriously questioning whether Atlassian has the will to support custom domains in their cloud suite. I find their replies to be ridiculous and unreliable, but I'm going to take their word for it that they're at least working on it.

Let me start by explaining why a custom domain is needed and how it's been done so far from the perspective of a Confluence Cloud user. Confluence started as a server version, which naturally has the freedom to set up a custom domain. Organizations I've worked with that use Confluence on an intranet domain and depending on the usage scenario, they either put Confluence on the internet to provide information, or they use Confluence as a replacement for a website, as I do now, or they use it as an FAQ database.

In these usage scenarios, a custom domain is important. Customers often judge the credibility of the website they're visiting by the domain it's being served on. We recently saw one mastodon community block an entire TLD because it was a free domain. Custom domains are an important feature, at least from a conversion standpoint. When Atlassian stopped selling and supporting the server version and supported the migration to the cloud version, we had a dedicated team to help us with all the weird things that happened during the migration, but almost exclusively custom domains were not supported at all. So customers who moved to the cloud could no longer use their custom domains, which caused the trust issues described above.

In recent years, the number of services that are enterprise-grade and don't support custom domains has increased dramatically, and I briefly thought that custom domain support is the order of the day, but I'm going to be a little conservative and say that I think it's still a trust issue for certain organizations to use custom domains. Some services have come out that mitigate this situation, and for close to a year, this website used a service called cloak.ist to provide custom domains on a limited basis. (The reason it's in the past tense is that we no longer use it, but that's a story for another time.)

Now let's talk about why I think Atlassian will probably never offer custom domains with their cloud product, at least not forever. Atlassian has ended sales and support of the server version of Confluence. The server version had been a reasonable choice for small to medium-sized businesses to meet their security needs. For a modest license fee, you could get a very good enterprise resource planning tool. However, it seems that customers who were using the server version were not renewing their licenses as frequently as Atlassian would have liked. Several organizations that were previously using Confluence were also running versions that were several years behind the latest version. None of them were using the latest version, and this gave Confluence a bit of an antiquated image in the minds of their users (employees).

For Atlassian, the way out of this situation is to get rid of the server version for small and medium-sized businesses and move them completely to the cloud. First of all, small and medium-sized businesses probably don't need the same security as the data center version, and secondly, once they're in the cloud, it's much easier to apply version updates and new licenses. The price increase for the cloud version that happened after the server version's end of support is a pretty good indication that this plan works. The cloud version is pretty easy to manage and quite modern from a user (employee) perspective.

To keep it that way, the feature that has appeared most often on our roadmap over the last couple of years is compliance with various compliance requirements, and since we've brought customers who were using the server version to the cloud, we've had to satisfy them ourselves, and we've spent a lot of resources on that. Now we're almost at the end, and once the data residency change feature is released to all instances, the plan to migrate the server version to the cloud version is almost finalized.

At this point, some of our larger customers, not just small and medium-sized businesses, may still not want the cloud version. They still need stronger security and may have requirements that are hard to consider for a smaller business (in fact, it's hard to imagine what the requirements would be at this scale), so we've left them the option to stay on the server, which is the minimum contract unit for a datacenter license of 500 users. If you look at the Atlassian cloud roadmap, there was a scale-up to 35,000 in the latest Confluence update, and I think Atlassian decided that it was okay to let customers of that size still use Confluence in an on-premises environment.

On the other hand, all the businesses smaller than that have consolidated to the cloud, eliminating the need to maintain servers because they don't have to comply with the licensing that they initially said they would, and because the cloud offers a lot of compliance anyway. In this situation, I think the CLOUD-6999 requirement could be seen as a feature that we don't have a reason to offer to our small and medium-sized customers who are not data center customers.

Data center customers are irrelevant to that requirement. They already have the same functionality naturally available to them whether that requirement is developed or not, because they are already serving in an on-premises environment. They're already paying enough for Atlassian. So I'm a little bit annoyed, but this thought process leads me to believe that Atlassian is not going to make CLOUD-6999 available in the cloud forever.