Atlassian, we need to talk about Data Center

Rodney N., ,

So – I didn’t want to write this post. I consider myself an Atlassian Fan, and in general, I really like the company. And honestly, I’m a bit nervous about writing this piece, as I don’t want to bite the hand that feeds me. I have a great career as an Atlassian Admin, and I tend to work well with the company.

But during last week’s Team ’22, I couldn’t help but notice something. Well, it was actually a lack of something. There was not one mention anywhere of Atlassian’s Data Center Products. Zero, zilch, zip. And that bothers me…a lot.

You see, way back in November 2020, I had an interview with Atlassian’s Harsh Jawharkar (who, it should be said, is no longer with Atlassian). In that interview, he said, and I quote:

“We are committed to investing in Data Center for the long term, and we hope that the velocity with which we are shipping new features and functionality to Data Center – and the fact that we are launching a brand new Data Center product, Bamboo Data Center – reassures you of this commitment.”

Harsh Jawharkar, Nov 2020

I published that in good faith, and I have advised multiple companies on their large Data Center instances since. I honestly felt I could help guide companies confidently on that platform and deployment model because Atlassian would be there with it. Surely they wouldn’t ax both On-Premise models.  

But lately, I’ve noticed two things that make me think otherwise. First, despite Jira 9.0 is on the horizon, Atlassian has only released a few Community posts about it. 

And second, they got zero time during Atlassian’s main conference. Given those factors, I have to add my name to the list of people who do not think Atlassian will support Data Center for much longer. 

“But Atlassian Cloud is where the growth is.”

I hear this often from Atlassian. It’s how Atlassian justifies having such a disproportionate amount of their Development time go to Cloud. However, I’ve come to think this argument is made in bad faith. Why?

Just look at Atlassian’s marketing efforts. As an Atlassian Admin, I am bombarded on the web, social media, and even in mobile games with ads for Atlassian products. And do you know what each and every ad I see says to me?

  • See how your company can perform on Atlassian Cloud.
  • Migrate to the Atlassian Cloud
  • Have you looked at Atlassian Cloud
  • Seriously, Atlassian Cloud is right there.
  • LOOK AT ATLASSIAN CLOUD NOW!

You load up Atlassian’s website, and Atlassian Cloud has a massive banner ad at the top. If I want ANY information on Jira Data Center from that page, I have to go to the pricing tab and click another tab under that page. 

So yes, You are going to see more growth on Atlassian Cloud. If you tell no one that Data Center exists, then no one will pick it. This situation is the literal definition of a self-fulfilling prophecy. I still think if given an even marketplace and chance, customers would pick Data Center over Cloud most of the time. And I think Atlassian knows this, hence why they stack the Marketing deck so heavily in Cloud’s favor. It’s the direction they want to move in, and they are doing anything they can to see that future through.

Don’t #@!% the customer

So this brings me to my next point. We’ve established that given their development time and marketing, Atlassian hugely favors Cloud over Data Center. So comes the next logical question, “Why?” To understand this, I think we need to establish a brief history of how Licensing has worked for Atlassian in the past. 

On the original Server line of products, when you first purchased Jira, Confluence, or any of their other tools, you were buying a perpetual license. If you never gave Atlassian another dime, you could still use your products for so long as you can keep a server running. This model has worked well for some companies, with Mojang coming to mind as a great example. You buy a Minecraft license, and you are good for all future game versions. Mojang funds all new development off further sales of the game and licensing of their IP. This arrangement has worked out well for them, considering the Microsoft buyout.

However, for server products, there is a wrinkle. You can only upgrade so long as you have an active support contract with them. This stipulation is how Atlassian guaranteed the continued funding they’d need for their development efforts. Most customers were happy to pay the annual fee, which was half of a new license plus half the cost of whatever Add-ons you were running. 

But do you know what’s better than half the cost of a new license every year? All of the cost of a new license every year. When Atlassian released its Data Center line, it came with an Annual Subscription. This meant that you didn’t get to use the products if you did not renew the subscription and pay for Jira, Confluence, or Bitbucket every year. And Atlassian pushed HARD for people to move from Server to Data Center.  

If this sounds familiar, you can stop me, but Atlassian started by putting features only in the Data Center. But even then, it was still more expensive to run Data Center over Server, so not everyone switched. Then they turned up the marketing on Data Center and barely mentioned the Server options. But some people still held out. Finally, they announced the discontinuation of the Server line of products to force those final holdouts off of their perpetual licenses. This situation is where we are now. 

The thing is, there is a solid argument that the Data Center model is superior to the Server product. You can scale it up as you need to; it has all the features and capabilities of Server, and if you aren’t ready for the full multi-server setup, the migration to Data Center is still as simple as pasting a new License code to your Server instance. 

However, do you know what’s even better than all of the cost of a new license every year? All of the license cost every month. This is where their Cloud product comes in.  

When the Server sunset was first announced, I argued that moving to Cloud *could* be less expensive than running Server. And I stand by the math I used then. But unfortunately, my math used discounts that are no longer available. So as it stands now, at most comparable scales, I cannot get the Cloud product to be cheaper than my Data Center installs – and that’s even before I consider Apps for Cloud, and after I consider that Atlassian has been aggressively increasing their price here lately.

However, unlike the situation between Server and Data Center – a Data Center migration to Cloud is no small undertaking. The base architecture is entirely different, meaning you won’t have the same capabilities. So if your teams built integrations based on those Capabilities, they have to be completely re-engineered. In addition, if your teams rely on certain Apps for your Atlassian Stack, those might not be available – and if they are, they might not have the same feature set. This means you are going to have potentially difficult conversations with your teams. No – Atlassian worked hard to make Data Center a drop-in replacement for Server. I don’t see how they will always be able to make Cloud a drop-in replacement for Data Center – especially for large instances.

Open company, no bullshit

All this would be fine if it wasn’t for the fact that Atlassian is still publically claiming they plan to have long-term support for Data Center. I mean, a corporation will do what they do, which most of the time is extract the most money for the least amount of effort.

But people are advising their teams, management, and companies based on Atlassian’s guidance about Data Center. Based on this guidance, I’m advising other Jira Admins about their careers. This is never a position I thought I’d be in, but here we are. And what happens when we take Atlassian on their word – despite every indication – and Atlassian pulls the rug out from under us (again)? We’ll be worse off for an eventual migration because we didn’t take the time to prepare we could have. If I’m being honest, THIS is what bothers me most. 

Honestly, I’ve started preparing my internal teams for this. I’m just upfront, saying I don’t trust Atlassian long-term to provide us with a “No Bullshit” answer on Data Center and that we should start looking at what a Cloud Migration could look like. In the meantime, I’m advising Jira Admins to get Cloud Ready and Cloud Certified. It’s what I’m planning to do long-term.  

Don’t get me wrong; I’m not running for the exits yet. Atlassian otherwise has been a fantastic company to work with, and I still see my long-term future with them. But I think it is on us as Customers, Partners, and Influencers to hold the company to the standard they have set for themselves. If you, Atlassian, have plans now to discontinue Data Center, be Open with us, and don’t feed us Bullshit. 

What do you think?

Do you think Atlassian has long-term plans for Data Center? Let me know if you agree and why or why not? I’m looking forward to a lively discussion on this and can’t wait to read your comments.

Yes, I still owe you a keynote recap. I’ve not been feeling the best this week, and thus my writing time has diminished. But I am working on it, I promise! Now that I have this conversation out of my head, I can focus on it fully, so that should help. 

As always, you can find all my links on Linktree. I always enjoy reading your comments, and liking the posts helps more people discover the blog!

And don’t forget, you can subscribe below to the blog to get an email every time we publish a new post. It’s one of the best ways to make sure you never miss a new post.  

But until next time, my name is Rodney, asking, “Have you updated your Jira instance today?”

Published by Rodney N.

Rodney has been working as a JIRA and Confluence Administrator and consultant for 5 years, and holds his Atlassian Certified Professional Cert for JIRA Administration and Atlassian System Administration.

22 Comments

  3. I am working as an Atlassian professional over a decade and I am a Fan through and throug, when it comes to Server or Datacenter. There are no customer requirements that are, if reasonable and though through, not doable with these products. Either with AddOns or own scipts and self written addons.

    But Cloud? Hell no!

    The Performance is bad, the features and design are horrible. How often I had to tell my customers “this is not working in cloud”, “nope, you can’t do that anymore”, “Only some features are available”, “nope you can’t change the design back to the original”. Or even that features were enabled without me or the customer noticing and I am searching like a maniac for a menu item, because it is now named “people” rathere than “users and roles” and is even located in a different spot… this just sucks!

    I have customer that have large datacenters with up to 40k users running. Fully optimized and customized to their needs. I don’t see those customers moving to cloud anytime soon. If Atlassian decides to cancel DataCenter I can assure that those customer will look for other alternatives as they are, also becuase of compliance reasons, not able to migrate into cloud.

    And right now, if Atlassian even decides to scrap Datacenter, I would leave my Atlassian profession after all this years.

    Like

    Reply

  4. A scary aticle for so many reasons, and I have always anticipated that Datacenter would it, but predicted this 5-10 years away, – but it may be closer. As the publication of Your aticle, part of Atlassians Cloud has been down for +7 days, do You think this will change much? I dont – and we forget so easily…. I am actually becoming more and more of a fan of the Cloud – mainly due to its many features that Server/datacenter not have, they are genrally more or less stalled….. But speaking of the current outage of the Cloud – If You have a proper offline backup (proper as in best effort, as the current Cloud backup is somewhat poor and ratelimited) – it should be possible to start on a new site/Url and restore You backup right? Not perfect, but perhaps better than waiting 2-3 weeks for the restorering from Atlassians side…. Anyways, Im in the midlle of migration several customers to the cloud, and even the smallest one it some of a hassle, so many things are not Migrated at all, and so many Addons gives a number of problems, and permissions in cloud are somewhat (for AddOns different) etc etc…..

    Liked by 1 person

    Reply

  5. Unless something changed recently, Atlassian didn’t offer BAA for companies to requiring HIPAA compliance. Jira Cloud is simply a non-starter for many companies without assurances on data location.

    Like

    Reply

  6. We had been using the small $10 server licenses for a while and were planning on shelling out cash for an upgrade before the announcement hit. Instead, we switched to FOSS solutions without the fear of having support dropped out from under us.

    Honestly, I think the reason Atlassian is trying to move to the cloud (without saying it) is probably an over-zealous attempt to stop piracy. It’s a little bit of an open secret that the entire Atlassian self-hosted suite (Server, Data-Center, and all the plugins) have been cr@cked.
    I believe that Atlassian is unable (or unwilling to dedicate the resources to) stopping this piracy. Instead, they would rather not admit their failure and instead, slowly turn up the heat on their loyal Server and Data Center customers because of flaws in *their* code. Data Center is doomed because their product has been cr@cked and Atlassian can’t or won’t fix it.

    Obviously, no Enterprise in their right mind would take such a risk (I say GTFO away from Atlassian now), but Atlassian put many a customer in a position to not be in their right mind. Especially when a minimum $40,000 a year price tag is attached to their only option.

    Like

    Reply

      1. There definitely are plenty of happily paying customers out there. That’s not what Atlassian sees, however. What they see is all the potential people out there who *should* be paying but aren’t. Whether it’s piracy or sitting on an old and insecure license, they want to put a stop to both of these. They are very likely making the common mistake of thinking that these are users who would pay for their product if they were given no choice, which isn’t true, and the legit customers get burned because of it.

        With Atlassian cloud, they don’t need to worry about either group. Either pay them regularly or have your data deleted in 30 days. There is no other option.

        Like

  7. Guess the plan is no plan for DC. Look at products like Bamboo, Fisheye / Crucible. They will have their own user-base for customers who can’t/will not move to cloud (i.e. Medical, Defense etc.). These users will pay their own premium for having ticketing system that will just provide the bare functionality. When the clientele is still – for a significant part – on DC/server they probably will not communicate the future of DC and or keep us in the dark.
    Also look at other companies who have similar roadmaps (i.e. Office 365 vs on-prem)

    Like

    Reply

  8. I’m out – I am using my Server License until it’s dead. Recemtly suffered a H2 Database corruption and had to replay the backup.
    But I will not trust my data into this black hole.

    The current outage should put Jira/Confluence on any one’s blacklist.

    Like

    Reply

  9. “Switch to Cloud” they encouraged. Now almost 2 weeks in with no service after a borked maintenance routine deleted customer instances (only some, but yes ours). Painful, and seriously damages the likelihood of getting the intended buy in for scaling this out widely in the organisation. I know it’s unlikely to happen again given the negative effect on their valuation, but trust is an important element of cloud adoption. It’s been an unexpected education.

    Like

    Reply

  10. Great article Rodney and well said! You’ve managed to spell out the track record of Atlassian and no one should really be surprised at where they’re headed. SAAS is the way most IT software has been going for sometime now. As a public company it’s in Atlassian’s best interests to move this way. The profits are better, as you mentioned. Will DC go away? Probably not anytime soon, but will the features stack up against the Cloud product over time? I’d say not. To that end, some companies will reluctantly stick with an inferior DC product or be forced into Cloud for more robust features. It’s wise to advise your teams to get trained up on Cloud.

    Like

    Reply

  11. Hi! I think your post is basically right, but I would add some nuance.

    First, “long-term” is a little ambiguous. How long do we really need or expect Atlassian to maintain their DataCenter offerings? 5 years? 10 years? 20 years? Certainly until the end-of-life date of any existing release goes to at least 2024, and we know that number pushes out further every time they issue a new release. Heck, even *Server* support isn’t fully ending until 2024, and Atlassian communicated about that extensively.

    Let’s recall another product that Atlassian has been slow-walking for *years*: Fisheye/Crucible. They put it into basic maintenance mode a few years ago (https://confluence.atlassian.com/fisheye/fisheye-and-crucible-are-in-basic-maintenance-mode-987143949.html). Yet since that announcement, they’ve shipped a bugfix release as recently as last month (https://jira.atlassian.com/projects/CRUC/versions/98296). Yes, putting a project on basic maintenance may be disappointing for its fans, but you can hardly say that Atlassian pulled the rug on their users. Given how relatively small the customer base was for Fisheye/Crucible, it’s hard to imagine Atlassian doing a rug pull for flagship products that bring in millions each month. They could, but we’re assuming they’re greedy, right? So, slow draw-down on DC? Sure, that’s actively underway. Rug pull? Nah.

    Speaking of greed… While I have no doubt that those sweet, sweet monthly subscription dollars are a major consideration, there are several other reasons that make sense of what Atlassian is doing. Consider that, as you pointed out, DataCenter products in general still have a decent amount of use cases and integrations that simply aren’t feasible or well-supported in Cloud. Of course, they and partner app developers are trying to close that gap, but that takes time, especially as their Cloud development platform remains a moving target. If Atlassian wanted to even get to feature & experience parity in Cloud, they would need to do exactly the kind of over-investment in Cloud, minimal investment in DC that they’re currently doing. Arguably, that should work out more or less fine, since DC had the features that made users want to jump on board anyway.

    A fellow commenter brought up GDPR. A big facet of the move to Cloud is, I think, regulatory compliance & risk management becomes more straightforward. No more, “Well, yes, it’s basically compliant, but if you do such and such with a plugin, or run your environment in this way, well, that’s on you.” Yes, Atlassian still have to make strides here, but the prospect for some companies to turn over a lot of regulatory risk to Atlassian is likely going to be too good to pass up.

    I say all this as a guy who has been making his living for the past 6 years writing plugins for the Atlassian on prem environment. I like it. I know what its warts are, but I’m productive there. I know the end is coming, but I’d say it’s coming slow enough that I’m not frightened for my career. I have time to upskill in other domains, and so does everyone else who is in some kind of partner role. I think you’re wise to tell your peers to be Cloud ready, but I wouldn’t fret too much. Our heads full of DC arcana will be lucrative assets for a few more years, at least.

    Sure, Atlassian have lived long enough to see themselves become the villain (https://news.ycombinator.com/item?id=11171465). But they’re more like one of those ambiguous, well-intentioned villains, like Xanatos from Gargoyles than they are like the Joker.

    Liked by 1 person

    Reply

  12. Cloud? No.

    Thanks for finally sharing an honest sentiment that many MANY Atlassian admins share.

    Atlassian @#$@#%@ the customer big time.

    – Very large company

    Liked by 1 person

    Reply
  13. Pingback: Atlassian Team 2022 - the most important news you need to know < Deviniti
  14. Pingback: So let’s talk about that Cloud Outage – The Jira Guy
  15. Pingback: Your Comments, Answered! – The Jira Guy

  16. We operate in a governmental space (vendor). its legislated that certain data isnt allowed to leave the borders of the country. there arent any AWS servers here either. so guess we gonna have to find another jira/bitbucket 😦 was looking to up our license users and got greeted with a “then you either go cloud or datacenter at $42000 a year” and renewal is same price as new. big mac index thats about 12 000 burgers (no fries). dont really know what decent alternatives there are 😦 no ways management will buy into this pricing..

    Like

    Reply

  17. Atlassian has completely lost the ball. They have lost most of our business with the deprecation of the server offerings – granted we are tiny fish, with 25, 50, 100 seat sites, but we work with big customers and teammates. Data Center is way out of scale for us. The ridiculous ‘simplification’ of Cloud, removal of essential functionality with the pretty but crappy new editor, make it unusable for many of our use cases. And as others in this thread mentioned, our compliance constraints require on-premises hosting.

    I’ve been an admin and evangelist for Confluence/Jira in our company for over a decade. I will not do any more with it other than supporting our existing sites until we replace most of them. We peaked out with an awesome collaborative space on a recent (ongoing) project using the excellent Comala Workflows for CM, the Scroll Exporter to spit out deliveries and custom Python automation of wiki content creation and data export from/to our engineering operations.

    All our new work and company self-hosted sites are transitioning to XWiki and Redmine. XWiki may be missing a few things but has a much better API and data model that make customization, automation, extension and integration much easier and more capable than Confluence. And it’s free as in beer unless you want to pay for some hand holding.

    The final nail has been the utterly abysmal lack of response to the “inputs” from folks like those of us on the community boards and the Jira issues sites for the Atlassian products. If they don’t give a ^&$%&%& about what their biggest (past) supporters say, they don’t deserve our business or support.

    Like

    Reply

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.