As it prepares to abandon its on-prem server products, Atlassian is content. Users? Not so much In October 2020, Atlassian announced that it would end support for its server products on February 15, 2024. With that deadline now less than five months away, the Australian developer is content it's done the right thing by customers – yet has warned investors the move is a risk to revenue. Atlassian offers its wares in three …
... and here in Europeland and in my company's line of work we have data souvereignty and strong privacy regulations. I sure hope the powers that be have a migration plan away from things like confluence and jira in mind.
The other problem is that coud is just other people's computers. Some in our management get that. I sure do hope they show Ye ancientte signe for banishing daemons (as Sir Pterry described it) and move to a different product (I know that the technically minded are pushing for it...)
It isn't my problem any more cos I left 2 years ago but my last company was the type where data security was very high up the list and putting things like software bug reports on a service where the data left the building was an ultimate no-no. We'd also integrated Jira and Confluence with several databases which made life a lot simpler and that is not something that could be replicated using the cloud offering.
The thing that drew us to Atlassian was the huge flexibility of the products and the plugins.
We became customers because you could download it and run it on local servers with no hassle. It saved time. The eventually got more dependent on it and will probably move to the cloud.
But we would not be customers in the first place if we had to have managers justify yearly costs upfront and do all the paperwork for having data off site.
More or less as soon as they announced they were withdrawing all but the DC license, we saw the writing on the wall and started planning to migrate away from JIRA and Confluence.
Ultimately, we ended up with on prem Gitlab - the revision control integration with tickets is far better than we had with JIRA (though, to be fair, our repos were in a non-Atlassian product) and the project wikis do what we needed from Confluence (the only thing is, a little bit of discipline is needed to ensure you don't end up with a plethora of Wikis to go hunting through).
If Gitlab ever rug-pull, we'll be able to move to the OSS version as a stop-gap (giving up a few features in the processes)
I'd be interested to know what Confluence is good for beyond wikis, or, as in the case of anything technical, ReST or Markdown files that sit in VCS. Exporting from Confluence is a particular nightmare with only two formats offered: PDF or Word.
I've currently got a project that has the repos on Github but uses Jira and Confluence for management and documentation, with some additional stuff scattered over Sharepoint. And no repo for the database schema!
At my last place we used confluence as the document master for an awful lot of customer and supplier facing documents. I can't remember which PDF exporter plugin we used but it had templates which allowed us to turn the page into a pretty nicely formatted document. This was done because the company had been bought numerous times over the years as well as some of the owning companies rebranding and as engineers always go with the 'copy and edit' method of creating a new doc we had some complete horror show word docs.
Some people don't understand that you don't change a document font by selecting all the text and picking a new font from the dropdown. Every time you started a new paragraph or added a bullet list it would switch back to an old font.
After this we only had to change the PDF templates when those above us decided to change the company logo and colours.... about 3 months after giving everyone branded mugs, lunch bags and jackets with the old logo... go figure.
Right, but the export step can't be the reason for choosing Confluence. Curently, moving a DOCX to ReST, so I can manage the damn thing over time and sphinx + rhino to create the PDF. Elsewhere the network ports are in Confluence… In another project, the developer documentation is in Confluence. Call me an old stick in the mud, but I text helps me concentrate on the content and quite a few editors provide the tools to make editing easier. Maybe if Confluence would deign to store stuff in ReST I wouldn't complain as much.
We used it as we had it. Confluence had been used for several years as an engineering wiki/collab tool and we previously had a site wide sharepoint that everyone could contribute to for formal docs. When IT moved to a new version of sharepoint they decided that only a very small number of people would have licenses so maintaining the already very badly mangled documents became an even bigger nightmare. IT and management didn't want to spend the money on more sharepoint licenses, we had confluence and we had the PDF exporter (as the built in one is a bit crap) so we did a 'two birds with one stone' migration coupled with 'path of least grief'.
It is not the way you'd go given a clean sheet of paper.
The number of docs we found where the author had just kept hitting enter to get to a new page....
There seems to be no consideration given to users where data security is about more than confidentiality.
For these users, a cloud instance is never going to be acceptable until there is a 100% guarantee that it is secure, encrypted and protected against loss.
Even if it's 100% guaranteed (which, of course, would be a lie) there is no chance of my employer agreeing to have source code outside of our own data centers. We rely on absolute security that we control.
No cloud. Ever.
We have roughly 1000 developers/testers/devops/product people in our organisation and we use Atlassian (Jira/Confluence) deployed in-house. Nice chunk of change for them about to disappear I imagine.
For these users, a cloud instance is never going to be acceptable until there is a 100% guarantee that it is secure, encrypted and protected against loss.
And for software that you host on-prem, you have a 100% guarantee that it is secure, encrypted and protected against loss?
Perhaps not, but the in-house people whose job it is to ensure that do in fact have skin in the game, as it's their livelihood that is at risk.
What's the absolute worst consequence to a cloud supplier if they "accidentally" let some miscreant take or alter their customers' data?
Unless and until the CEO of the cloud supplier is personally jointly and severally liable for all the direct and indirect consequences, including imprisonment, I don't see why they should be trusted with company crown jewels.
The cloud providers have a skin in the game too - if there's data loss due to their neglience, it's game over for their entire business.
In practice, data loss from cloud providers is generally down to the user's own misconfiguration. World-readable buckets for example, or services open to the Internet which shouldn't be. You can make the same sorts of admin mistakes on-prem too.
The cloud providers very likely have much better security teams, practices and procedures than you do.
perpetual, adjective: never ending or changing
Yeah, right.
"This was how every software product worked before the internet."
Another beancounter who knows shit about IT, but knows how to screw customers.
One was macroeconomic uncertainties.
I have to see a moment when there will be macroeconomic certainties.
Bean counters are the ones driving the dash to cloud. Because, bean counters never ever take account of possible eventualities such as "Amazon is broken today". They assume that cloud providers are some how infallible...
It's also bean counters who underinvest in on prem, quite often. I recall British Airways nearly going out of operations because a hot weather spell nearly took out their aged, creaking server building (their only one).
Anon as based on recent employer history.
They ditched on prem Atlassian products (Conflunece & Jira) and are using alternate solutions.
Not just cost, there are risks with cloud (e.g. outside of EU data processing issues, security in general (Atlassian have a dismal good history when it comes to nasty exploits, with on prem at least you have some decent control to stop exploits hitting the server (on prem instances were restricted access, only available to some internal network users, which also helped a lot in risk mitigation ))).
The general we don't care about on prem, new features pretty much cloud only, like it or lump it approach Atlassian took did not help either.
Bad choice as (in this particular use case, not unbreakable vendor lock in) plenty of alternatives for collaboration and bug / ticket handling software (and data migration was viable, though was a bit of a pain which could potentially dissuade some companies if lacking staff time / skills to do data migration).
I feel like opening and successfully closing three levels of parenthesis with a degree of elegance calls for some sort of award, similar to the recognition given to achiement in gymnastics. Most of us (and I include myself in this) are far too prone to throwing in an opening bracket, then forgetting to close it. Although programming (and using Excel formulae) does tend to reduce the error rate a bit.
《I feel like opening and successfully closing three levels of parenthesis with a degree of elegance calls for some sort of award》
LISP (or Scheme) doesn't get much of a nod here - and surely LISP etc would consume the lion's share of the global production of parentheses. :)
I ditched Jira when they took the tree out of sourceTree: https://jira.atlassian.com/browse/SRCTREEWIN-7176
Made a good product, proceeded to shitify & force release it, with core feature removed. Won't fix last major version that was pretty perfect (minus the CVE).
Then! They held their position as if angry developers were the enemy.
I honestly don't know how they are in business, the hostility they presented against overwhelming feedback told me all I need to know - there'd be no decent support, sane decisions coming out of there.
Since we had our on-prem go to their cloud, response for page loading has been dire. Complaints are across the department.
They run a serious risk of losing business unless they get this sorted ... pronto ... but in our experience, they're dragging their heels and it is a dismal service IMHO.
《I'd no more place my systems in the hands of a cloud operator than I'd place my john thomas in the hands of a lunatic with a pair of scissors.》
Loony Scissorhands is only going to remove the conductor not the orchestra - your cloud operator has your family jewels so at some point there is a real risk that you are only going to be competing against counter tenors.
Our corporate overlords at Cthullu International used on-prem Atlassian. They demanded we move off when they went cloudy - but moved us to cloudy Github ?
Wondered why we had a sudden rush to move to a cloud - when they are paranoid about having all our data in R'lyeh
It's all about the money !
Microsoft, of all companies, seems to have come up with what I consider a reasonable compromise on this front. There's their Office 365 version, which is where they focus their development efforts, and then every couple of years they roll up all the changes and release it as a stand-alone version like Office releases of yore. So, the cloudy versions of Atlassian's wares could be where features show up first, and then every so often they just take all those changes and release a new on-prem version. Both get security updates and other critical bugfixes, 'natch, but the brunt of the development effort can go to the cloudy version. The on-prem version being like a LTS release.
Of course, with the whole "500 users, take it or leave it" approach almost seems like they want to drive away non-cloudy customers. Seems kind of stupid IMO, but there's no law that I'm aware of, anywhere in the world, that says a business can't shoot itself in the foot and drive away customers potentially imperiling its financial solvency. I'm sure their competitors are all rushing to make sure they can import data from Atlassian software and are blasting out marketing emails saying how they offer on-prem software with more flexible licensing terms.
Not really the point, but considering how little it costs them to just say the version they distributed via Office 365 at the beginning of Q1 of every third year (or whatever criteria they use) gets packaged as a stand alone version, I think it'll be around for a good long while yet. Let's just say it costs them $100K every 3 years to update and distribute the stand alone version, and it makes them $1m in sales, which is probably a pretty low estimate. It's definitely worth the while to keep making it. You probably have one or two part-time developers working on backporting security updates, and that's basically it. Everything else piggybacks on Office 365.
The real cost though is that every standalone office costs them the $10-100/month they can charge for Office365
Unless they can charge $$$ for every new standalone and everyone renews every coupe of years and there's no piracy - they lose money however cheap the development
Earlier this year I was looking at using Jira, Confluence and bitbucket for a small team.
We have to have on premises or under our control due to security requirements from customers.
So I've gone with gitlab premium...
While the majority of the functionality is fine I can't quite get the Wiki"s to be as good as confluence. I'd love to see a tutorial or write up on how to achieve something similar or just a more feature rich wiki with the ability to export as a document.
In any defence or government project that uses atlassian products good luck trying to push them towards a cloud only option.
They will *NEVER* adopt SaaS/cloud if it means storing the work off premises. Any project requiring security clearance will never, ever get stored off prem.
We literally spent $2500 per user on Helix ALM because of the Atlassian encloudification. We've been quite happy with the move.
Jira was never that good to begin with. There have always been a countless number of options out there, almost all of them are better. If you're going to be forced cloudy, at least use something good, like CA/Broadcom Rally.
The day Atlassian announced the end of Server, I swore never to use an Atlassian product again and started migrating all existing data off Atlassian products onto open source alternatives. The risk of using open-source is at least manageable (as long as you don't risk anything built on PHP or MySQL, at least, as those tend to be architectural smells that reek of poor thinking elsewhere in the system as well). The risk of vendor lock-in isn't, for us.
So far, we're very happy.
Cloud is incredibly overrated. I've never really bought into the hype but people feel swayed easily. There are simply too many issues and I'm not talking about the Richard Stallman, right to privacy and digital libertarianism and everyone needs a server at home connected to symmetrical 10gbps internet arguments.
Maybe I need more real-world experience" with the "IT industry" my university lab tried MS planner. While useful tech the conversation soon turned to how do we backup and restore?
Which many of these platforms just don't get. Usually the only way to restore is via some unofficial third-party tools that require more money to pay for.
Then there is lead time for recovery. How long will it take to restore all the data, especially if bandwidth is limited?
Then there are data security aspects? Want to store some of our population cohorts in the cloud? Ya maybe its legal to do technically but if there is a data leak and my institute is sued, it still has to pay for the lawyers until it successfully defends and the plantiff deposits a refund on legal fees in their bank account and may still suffers major reputation damage.
Is it cheaper than a local HPC cluster? You quickly find out it isn't once you use it regularly. They might give you x amount time free on AWS/Azure but soon it costs just as much as buying the hardware yourself.
Despite all these pitfalls what I think sways people to use cloud software and ignore the risks for some things is popularity based on good user experience, simply because of the market position of these big tech companies enables them to hire lots and lots of designers and software developers, along with a bunch fancy marketers. Things smaller firms/startups can't unless they have been given tons of VC money to play with.
I largely agree except for on particular use case.
Cloud allows anybody to briefly access prodigious amounts of compute power for just a small amount of money. This is something that on prem cannot compete with. If one has the need to occasionally use vast amounts of compute, cloud wins.
It's a use case that the cloud providers probably aren't keen on really, because they'd prefer long running routine workloads that allows them to hone their capacity closer to the market average for maximum profit.
And apparently you can, as a cloud subscriber, run into the cloud's resource limits quite easily. I had heard that getting dev access to Amazon's GPU offering as a dev in the UK was difficult because the Europeans who wake up an hour earlier sign them out for the day first. They may have added more, since.
But yes, cloud is just someone else's computer, and the bespoke ways of programming them reek of vendor lock in.
The Cloudy version periodically gives your employees unannounced "distraction-free, heads down, working" days. No JIRA tickets or Confluence RFCs popping into your queue. Or take a break in the beautiful outdoors if you need documentation or specifications to get started.
Been trying to convince senior management that we're going to have to move off JIRA or move onto the cloud since I got the Atlassian email 4 years ago and they've punted the decision. We even made a plan, found alternatives, worked out what could be migrated, what we'd not bother with, what couldn't etc. But it just got bounced. Often cos there were more important things, sometimes because it was too far in the future to worry about.
While there might be data sovereignty issues with our clients, they were scared of the conversation so just didn't have it with them. I know we've not managed to sign off some cloud stuff in the past. Some clients won't care to be fair, but others are more sensitive and will. You don't know until you ask! They will all definitely care that we would be relying on an unsupported product, so staying on prem with JIRA etc won't wash. I quietly posted this story on the internal slack and now there's panic. I'm not panicking particularly because we still have enough time but someone above me is going to have to sign it off sharpish.
Thanks El Reg, you might have lit a fire under the suits! Anon for obvious reasons.
"Some users who contacted The Register told us they've started, and paused, migrations away from server products because the process was not easy."
It literally didn't work - we had to migrate twice because their systems weren't working correctly. Took weeks to fix.
Remember folks - it's not the cloud, it's just somebody else's computer.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
