Short answer: Atlassian stopped selling Confluence Data Center to new customers on March 30, 2026. Existing customers can still renew at the same size but can't expand after March 30, 2028, and on March 28, 2029 every instance goes read-only. Teams with large, heavily customized or regulated wikis should pick between Atlassian Cloud and a knowledge base they own now, while a calm migration is still possible.
All three milestones come from Atlassian's Data Center End of Life FAQ, which we checked on November 21, 2026. Each one is at 23:59 PST on the date shown.
Which products are affected. In scope: Confluence Data Center, Jira Software Data Center, Jira Service Management Data Center, Crowd Data Center, the Data Center mobile apps, and Marketplace apps for those products. Out of scope: Bitbucket Data Center and Bamboo Data Center, which continue past 2029 (existing customers get a Bitbucket Hybrid License covering Bitbucket DC, Bamboo DC and Bitbucket Cloud), and Jira Align. Atlassian says it will offer extended maintenance or multi-year extensions after March 28, 2029 "by exception only", with no published date. For FedRAMP High and IL5 it has made a commitment with no fixed date; Atlassian Government Cloud is authorized at FedRAMP Moderate. If Jira sits next to your wiki, the same calendar applies; our Jira Data Center end-of-life runbook covers the project side.
A wiki is a different migration problem from a ticket tracker. The content is mostly prose, but it is held together by page trees, macros, attachments, links between pages and a permission model people rarely document. Which option fits depends mostly on how much of that structure your teams actually rely on.
Atlassian's route for option A is the Confluence Cloud Migration Assistant, which Atlassian calls "the recommended way to migrate", supported by the Atlassian Migration Program. For many organizations that is the right answer and the least work. Atlassian also advertises a first-year incentive on Cloud Enterprise for which customers buying by June 2027 may be eligible (its offer, with its own terms), which is worth asking your account team about as one input to the Cloud number.
We don't show Data Center prices because Atlassian publishes none we could verify. Use your own invoice. The comparison has three lines:
The example below uses total yearly spend on the wiki as the comparison line because it is the number most IT leaders already have. Swap in your Cloud quote if that is the real alternative. Every input here is an example, not an Atlassian price and not a Ryz quote.
Illustrations only. Not Atlassian list prices; use your own contract and quote. The crossing month is build cost divided by the monthly difference between current spend and assumed run cost, counted from the first build month.
Two things move these lines more than anything else: how many custom macros and integrations must be rebuilt, and how many writers outside engineering need a rich editor. A docs-as-code setup for engineering content is cheap to run; a full replacement for every department is not. Put your own figures into the 3-year build vs buy TCO calculator and see whether the lines cross at all.
These steps apply to option C and to the owned half of the middle path. Most of them also make a Cloud migration go better, because they tell you what content you actually have.
List every space with its owner, page count, last edit date and attachment volume. Many estates have a long tail of spaces nobody has touched in years; decide now whether they move, get archived as static files, or are left to the read-only instance. Then count macro usage. Confluence stores pages in an XHTML-based storage format with custom elements for macros, so a script over exported page bodies can count which macros appear, where, and how often. Each Marketplace app macro is either rebuilt, replaced with plain content, or dropped. Exit check: a spreadsheet of spaces and a ranked macro list, each row with a decision.
Confluence Data Center offers PDF, Word, HTML and XML exports. The space XML export is a zip with pages, comments and attachments, and restricted pages stay restricted; it can only be imported into the same or a later Confluence version, and it leaves out Team Calendars. The HTML export holds pages and attachments, but only pages the exporting user can view, so run it with an account that can see everything. Exporting a space needs the Export Space permission, and only one XML backup runs at a time, so schedule large spaces. For a repeatable pipeline, use the REST API instead: GET /rest/api/content/{id}?expand=body.storage returns a page body in storage format, /rest/api/content/{id}/child/attachment lists its files with a download link for each, and the ancestors field gives you the page tree. Results are paged with start and limit; keep requesting until the response no longer carries a next link. Exit check: page and attachment counts per space match between the source and your export store.
Whatever stores the content, the model needs spaces, pages with a parent pointer and sort order, page versions, attachments with their own versions, labels, comments, and links between pages. Confluence keeps a version for every change and lets users compare and restore them; decide how much history you migrate (all versions, the last few, or only current) because it drives storage and runtime. In PostgreSQL, a pages table with a parent id and a versions table is a sound start; attachments go to object storage with checksums in the database. For engineering spaces, the docs-as-code alternative is Markdown files in a Git repository, reviewed by pull request and built into a static site by CI, so history lives in Git. Exit check: one real space converted end to end, with rendering reviewed by its owners.
Convert storage format to your target (Markdown or HTML) with a tested converter: rewrite attachment references, which the storage format records by file name, into object-storage URLs, and turn internal page links into stable ids so moves don't break them. Rebuild the few macros that carry real work (status tables fed from other systems, include-page patterns, code blocks, diagrams) as components. Search is the feature people notice first: plan full-text search over titles and bodies, filtered by permission, before anything else. If you want answers over the knowledge base, add retrieval on top of that index later. Exit check: a broken-link report at zero and macro output matched against the source pages.
Confluence space permissions cover viewing, adding and deleting pages, attachments, comments and restrictions, plus Export space and Admin. They can be granted to users or groups and add up across groups, and page restrictions for viewing or editing sit on top of them. Export the effective permissions per space and per restricted page, map them to roles from your identity provider, and test with real accounts that restricted pages stay hidden, in search results too. If Confluence authenticates through Crowd Data Center, note that Crowd is also in end-of-life scope, so the identity plan is part of this project rather than a follow-up. Exit check: a permission test suite that passes for every restricted page sampled.
Freeze the spaces that move, or sync deltas from the API nightly, and run both systems for a defined period. Track which pages people open in the new system and which searches return nothing; failed searches tell you where conversion or permissions are wrong. Keep edits in one place at a time per space to avoid split versions. Exit check: owners sign off space by space.
Redirect old page URLs to their new locations so links in tickets, code comments and chat keep working. Take a final XML export of every space as an archive, then retire the instance on your own schedule, well before March 28, 2029. After that date the read-only instance gets no fixes and Atlassian advises against leaving it on the internet, so a frozen archive you control is the safer record. Exit check: redirect coverage for the most visited pages, and a restore test of the archive.
Our AI pod teams use coding assistants for the repetitive parts of a wiki migration: storage-format converters, link rewriting, export scripts and test generation. Every change still goes through human review and tests before it touches your content.
A knowledge-base move needs a tech lead who owns the content model and cutover plan, one to three senior engineers for conversion, search and the editor, a data engineer when attachment volume or history is large, and part-time QA for permission and link testing. Many teams begin with a pod of 3 to 5 senior engineers (a tech lead plus engineers), and a full pod is about seven. A docs-as-code move for engineering spaces alone can start with a tech lead plus one or two engineers. Ryz engineers on your team start from $7,000 per month (our published range is $7,000–$15,000 per engineer per month, mostly mid-level to senior), quoted per team. Once you say go, engineers can typically start in under two weeks. Ryz engineers work on your team, reporting to your leads. Bring your space inventory and scope your team with us, or see the kinds of production systems our pods have shipped in our case studies.
Ryz is the wrong choice if you want a strategy-only engagement or a ready-made platform; for a standard wiki moving to Cloud, the Migration Assistant and Atlassian's partners are the shorter route.
Yes, at the same size. After March 30, 2028 existing customers can't buy new subscriptions, expansions, upgrades or new Marketplace Data Center apps, but they can renew for the users and apps they already have, or downgrade, until March 28, 2029. A 12-month renewal that would run past that date is prorated to it, and on that date the instance becomes read-only.
No. Bitbucket Data Center and Bamboo Data Center are excluded and continue beyond 2029, and existing customers get a Bitbucket Hybrid License that covers Bitbucket DC, Bamboo DC and Bitbucket Cloud. Jira Align is also excluded. Confluence, Jira Software, Jira Service Management and Crowd Data Center, the mobile apps and their Marketplace apps are in scope.
Confluence Data Center exports spaces as XML (pages, comments and attachments, with restrictions kept) and HTML (pages and attachments the exporting user can view), and single pages as PDF or Word. For a repeatable migration, the REST API returns each page body in storage format, lists attachments with download links, and exposes the page tree through ancestors. Most teams take an XML archive of every space and run the API pipeline for the content they convert.
It depends on the number of spaces you move, how many custom macros must be rebuilt, how much version history you keep, how complex your page restrictions are, and how long owners want the parallel run. A docs-as-code move for engineering spaces is much shorter than a company-wide wiki with many apps. We scope it from your inventory rather than quoting a duration up front.
The build is engineers times months times the monthly rate, from $7,000 per engineer per month, plus a yearly run cost for hosting, attachment storage, search and maintenance time. Compare three years of that against three years of your Cloud quote or your current Data Center spend, remembering that staying on Data Center only lasts until March 28, 2029.
Questions we didn't answer? Email info@ryzlabs.com.
Thanks — your message has been sent. We’ll get back to you soon.
Something went wrong while sending your message. Please try again or email info@ryzlabs.com.